OpenAI’s control structure, 2015–2026: who held which control right, and when →
Institutional dossier
The documented history of OpenAI: chronology, people, research, capital and compute, and governance. Every statement opens its source and its limits.
Records dated 2011 to 2026-10-29 · 71 undated
747 claims · 274 sources. The counts describe this selection. They do not measure completeness or independent confirmation.
Scope, evidence and limits
Description: An AI research and deployment organisation.
The records are grouped automatically. Each source passage shows how it was checked.
The dossier is computed from existing records. It is not a continuous history of the institution, and a missing record does not mean that an event or relationship did not happen. The chronology uses the dates of events and reports, not the date the public could first have known of them. Later sources can describe earlier events.
Select any statement to see its exact claim, qualification, date note, evidence passage, location in the source, source version and review status. The description above is Selekoa’s stored description of the institution, based on these sources:
Connected institutional dossiers
Institutions linked by a documented relationship. Each keeps its own chronology and evidence. When units combine, their earlier records stay separate.
- Open io Products, Inc. dossier →OpenAI stated in an update dated 9 July 2025 that the io Products, Inc. team had officially merged with OpenAI, as announced on 21 May 2025. Inspect evidence ↗
- Open Statsig dossier →OpenAI announced on 2 September 2025 that it was acquiring Statsig, subject to customary closing conditions including regulatory approval. Inspect evidence ↗
Chronology
Records about the institution itself and documented changes to its frameworks, by year. Expand a year to read the claims. People’s other institutional records and related documents have their own sections. Where a date gives only the year or month, the order within it is unknown.
2015Record year · 15 claims
- OpenAI’s launch announcement names Durk Kingma among its founding members.Durk Kingma → co-founded → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
The predicate means participation in founding; the source wording is founding member, not a statement of legal founder status.
Founding participation documented by the December 2015 launch announcement; year precision does not assert an incorporation date.
- OpenAI’s launch announcement names John Schulman among its founding members.John Schulman → co-founded → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
The predicate means participation in founding; the source wording is founding member, not a statement of legal founder status.
Founding participation documented by the December 2015 launch announcement; year precision does not assert an incorporation date.
- Andrej Karpathy’s career timeline places his OpenAI research-scientist and founding-member work in 2015–2017.Andrej Karpathy → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Retrospective first-person year-precision account; not a legal incorporation claim.
Retrospective career account supplies these boundaries at the stored precision; no exact joining or departure day inferred.
- Durk Kingma’s biography dates his OpenAI Research Scientist role to 2015–2018.Durk Kingma → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Retrospective year-precision role interval; founding participation is a separate claim.
Retrospective career account supplies these boundaries at the stored precision; no exact joining or departure day inferred.
- OpenAI’s launch announcement of 11 December 2015 names Andrej Karpathy among the group’s founding members.Andrej Karpathy → co-founded → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Read through a reader relay
Qualification and date notes
The predicate means participation in founding; the source wording is “founding members”, a group of research engineers and scientists, not a statement of legal founder status.
Founding participation documented by the December 2015 launch announcement; year precision does not assert an incorporation date.
- OpenAI stated in its recapitalization announcement of 28 October 2025 that it was founded in 2015 as a nonprofit.OpenAI → is stated to have been founded in → 2015, as a nonprofit Inspect claim & evidence ↗
Record about this institution · Stated in source · Supported Read through a reader relay
Qualification and date notes
A retrospective statement by the organisation. It does not give an incorporation date or name the founders.
Founding year as the source states it (year precision).
- OpenAI’s launch announcement of 11 December 2015 names Pamela Vagata among the group’s founding members.Pamela Vagata → co-founded → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Read through a reader relay
Qualification and date notes
The predicate means participation in founding; the source wording is “founding members”, a group of research engineers and scientists, not a statement of legal founder status.
Founding participation documented by the December 2015 launch announcement; year precision does not assert an incorporation date.
- OpenAI’s launch announcement of 11 December 2015 names Trevor Blackwell among the group’s founding members.Trevor Blackwell → co-founded → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Read through a reader relay
Qualification and date notes
The predicate means participation in founding; the source wording is “founding members”, a group of research engineers and scientists, not a statement of legal founder status.
Founding participation documented by the December 2015 launch announcement; year precision does not assert an incorporation date.
- OpenAI’s launch announcement of 11 December 2015 names Vicki Cheung among the group’s founding members.Vicki Cheung → co-founded → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Read through a reader relay
Qualification and date notes
The predicate means participation in founding; the source wording is “founding members”, a group of research engineers and scientists, not a statement of legal founder status.
Founding participation documented by the December 2015 launch announcement; year precision does not assert an incorporation date.
- OpenAI’s launch announcement names Wojciech Zaremba among its founding members.Wojciech Zaremba → co-founded → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
The predicate means participation in founding; the source wording is founding member, not a statement of legal founder status.
Founding participation documented by the December 2015 launch announcement; year precision does not assert an incorporation date.
- OpenAI’s launch announcement names Sam Altman as a co-chair.Sam Altman → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
This is a historical role snapshot. Neither the appointment date nor the end of the role is established by this record.
The date identifies this documented announcement or publication; it does not establish a continuing state.
- OpenAI published its launch announcement describing a nonprofit AI research company.OpenAI → published → Introducing OpenAI (2015)Inspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Public introduction, not a legal incorporation date. The description belongs to this 2015 announcement.
Date on the inspected publication; not the source-access date or proof of implementation.
- OpenAI’s launch announcement identifies Greg Brockman as CTO.Greg Brockman → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Historical title snapshot, not a current role.
Role snapshot at the dated source; no start, end or continuing appointment inferred.
- OpenAI’s launch announcement names Elon Musk as a co-chair.Elon Musk → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
This is a historical role snapshot. Neither the appointment date nor the end of the role is established by this record.
The date identifies this documented announcement or publication; it does not establish a continuing state.
- OpenAI’s launch announcement names Ilya Sutskever as research director.Ilya Sutskever → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
This is a historical role snapshot. Neither the appointment date nor the end of the role is established by this record.
The date identifies this documented announcement or publication; it does not establish a continuing state.
2016Record year · 3 claims
- OpenAI announced on 26 April 2016 that Pieter Abbeel was taking leave from UC Berkeley to work with OpenAI full-time.Pieter Abbeel → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Read through a reader relay
Qualification and date notes
The announcement gives no title, start date or end date.
Announcement date; not a start date.
- OpenAI announced on 26 April 2016 that Shivon Zilis was becoming an official advisor to OpenAI.Shivon Zilis → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Read through a reader relay
Qualification and date notes
An advisory role as announced; no end date is recorded.
Announcement date; not a start date.
- OpenAI published the Gym beta announcement, introducing a toolkit for comparing reinforcement-learning algorithms.OpenAI → published → OpenAI Gym Beta (2016)Inspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Publisher account of its own work; publication does not independently establish implementation, effectiveness or priority.
Date on the inspected publication; not the source-access date or proof of implementation.
2017Record year · 3 claims
- DeepMind’s human-feedback account credits its research collaboration with OpenAI.Learning through human feedback — DeepMind account (2017) → explicitly credits → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Specific attributed collaboration; not an inferred influence edge or transfer of people.
Date on the inspected publication; not the source-access date or proof of implementation.
- The first version of Deep reinforcement learning from human preferences lists Paul F Christiano with an OpenAI affiliation.Paul Christiano → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
A printed author affiliation does not establish employment, a job title, a start date or tenure, and says nothing about any earlier or concurrent affiliation.
Snapshot of the affiliation printed in arXiv v1 dated 2017-06-12; not an employment start date.
- OpenAI published its account of learning reward functions from human comparisons.OpenAI → published → Learning from human preferences — OpenAI account (2017)Inspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Publisher account of its own work; publication does not independently establish implementation, effectiveness or priority.
Date on the inspected publication; not the source-access date or proof of implementation.
2018Record year · 1 claims
- OpenAI announced Elon Musk’s departure from its board in February 2018.Elon Musk → had departure announced from → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Board role only. The announcement also envisages continued advice and donations, so this is not an end to every association.
Date of the public board-departure announcement; no general employment-end date is inferred.
2019Record year · 5 claims
- OpenAI published its announcement creating a capped-profit partnership, OpenAI LP.OpenAI → published → OpenAI LP — capped-profit structure announcement (2019)Inspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Publisher structural announcement, not independent verification of legal instruments. OpenAI is the umbrella atlas identity; the LP is not equated with the nonprofit.
Date on the inspected publication; not the source-access date or proof of implementation.
- OpenAI announced on 11 March 2019 that it was creating OpenAI LP as a hybrid of a for-profit and a nonprofit, which it called a “capped-profit” company. Investors and employees could receive a capped return, with the first round of investors capped at 100 times their investment, and returns beyond the cap would go to OpenAI Nonprofit.OpenAI → formed → OpenAI LPInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
OpenAI’s own account of a structure it was creating. It is not evidence of the partnership’s formation date, its agreements or any payment of returns. The post says that “OpenAI” would from then on refer to OpenAI LP and the original entity to “OpenAI Nonprofit”, and that caps for later rounds would be negotiated per limited partner and expected to be lower. Control is recorded separately (the record of the March 2019 announcement on board control); the umbrella OpenAI identity is the subject because the post speaks for “OpenAI”.
Date of the announcement (day precision), shown on the post and on the Internet Archive capture of 2019-03-11. The partnership’s own formation date is not stated in the post and is not recorded.
- Microsoft announced a $1 billion investment in OpenAI.Microsoft → announced investment in → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Stage: announced · exact USD 1,000,000,000 · transaction
Qualification and date notes
Announcement amount; cash disbursement, instrument and ownership share are not established.
Announcement date; the transaction stage is separately recorded.
- OpenAI announced it would move its services to Microsoft Azure.OpenAI → selected cloud provider → MicrosoftInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Stage: announced
Qualification and date notes
Azure is represented through Microsoft. A migration announcement does not establish completed migration or present exclusivity.
Announcement date; the transaction stage is separately recorded.
- OpenAI published its account of simulation-trained reinforcement learning for robotic manipulation.OpenAI → published → Solving Rubik’s Cube with a robot hand (2019)Inspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Publisher account of its own work; publication does not independently establish implementation, effectiveness or priority.
Date on the inspected publication; not the source-access date or proof of implementation.
2020Record year · 5 claims
- OpenAI published its institutional release of Scaling Laws for Neural Language Models.OpenAI → published → Scaling Laws for Neural Language ModelsInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Institutional release date; the paper’s earlier arXiv version date remains separate. Existing paper citations and method-use records are reused.
Date on the inspected publication; not the source-access date or proof of implementation.
- The first version of Scaling Laws for Neural Language Models lists Jared Kaplan with an OpenAI affiliation.Jared Kaplan → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Printed affiliation does not establish employee/contractor status or tenure. Johns Hopkins University is also listed; affiliation is not exclusive.
Snapshot of the affiliation printed in arXiv v1 dated 2020-01-23; not an employment start date.
- The first version of Scaling Laws for Neural Language Models lists Sam McCandlish with an OpenAI affiliation.Sam McCandlish → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Printed affiliation does not establish employee/contractor status or tenure.
Snapshot of the affiliation printed in arXiv v1 dated 2020-01-23; not an employment start date.
- OpenAI announced that Dario Amodei was leaving the company.Dario Amodei → had departure announced from → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
An announced departure, not an exact employment boundary. The unnamed new project is not equated with Anthropic in this record.
Public announcement date. The body mentions an employee announcement earlier that month; the exact final working day is not given.
- OpenAI identified Dario Amodei as VP of Research in its December 2020 departure announcement.Dario Amodei → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Role identified at departure announcement; neither appointment date nor final working day is established.
Snapshot at the dated source announcement; no employment interval or current appointment is inferred.
2021Record year · 1 claims
- The Codex paper (arXiv v1, 7 July 2021) lists Lukasz Kaiser with an OpenAI affiliation.Łukasz Kaiser → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
An author affiliation; it does not establish when he joined OpenAI, a title or tenure.
arXiv v1 submission date.
2022Record year · 5 claims
- Barnes described leaving OpenAI in summer 2022.Beth Barnes → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Retrospective participant account of prior employment, not an independently verified last working day.
The interview specifies summer 2022; the schema retains year precision rather than inventing a day. Start unknown.
- John Schulman’s biography dates his OpenAI post-training co-lead role to 2022–2024.John Schulman → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Retrospective first-person account; end-year precision is not an exact departure date.
Retrospective career account supplies these boundaries at the stored precision; no exact joining or departure day inferred.
- OpenAI announced on 5 May 2022 that Greg Brockman was becoming President, a new role.Greg Brockman → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Read through a reader relay
Qualification and date notes
The announcement does not give an exact start date or establish a continuing role.
Announcement date; not a start date.
- OpenAI announced on 5 May 2022 that Brad Lightcap would become Chief Operating Officer.Brad Lightcap → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported AnnouncedRead through a reader relay
Activity: announced
Qualification and date notes
An announced appointment. OpenAI’s March 2025 leadership update refers to him as Chief Operating Officer with an expanded remit.
Announcement date; not a start date.
- OpenAI announced on 5 May 2022 that Mira Murati was taking on the role of Chief Technology Officer.Mira Murati → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Read through a reader relay
Qualification and date notes
The announcement does not give an exact start date. It says she had led research, product and partnership functions over the previous 18 months.
Announcement date; not a start date.
2023Record year · 18 claims
- Andrej Karpathy describes returning to OpenAI in 2023–2024 to build a midtraining and synthetic-data team.Andrej Karpathy → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Role description follows the source; no formal title, transfer edge or current appointment inferred.
Retrospective career account supplies these boundaries at the stored precision; no exact joining or departure day inferred.
- Microsoft announced a multiyear, multibillion-dollar investment in OpenAI.Microsoft → announced investment in → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Stage: announced
Qualification and date notes
The announcement does not disclose an exact amount or realised disbursements; none are inferred.
The date identifies this documented announcement or publication; it does not establish a continuing state.
- Microsoft’s partnership announcement described Azure as OpenAI’s exclusive cloud provider.OpenAI → selected cloud provider → MicrosoftInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Stage: announced
Qualification and date notes
Azure is represented through its Microsoft counterparty. This describes the January 2023 announcement, not present-day exclusivity or an amount spent.
The date identifies this documented announcement or publication; it does not establish a continuing state.
- OpenAI is the author named on the GPT-4 Technical Report (arXiv v1, 2023).OpenAI → published → GPT-4 Technical ReportInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
The report names OpenAI as its author, with a contributions list; the arXiv submission is the dated expression.
Date of submission (an arXiv version or a journal’s received date, as the source states it); not a later revision or the venue’s publication date.
- OpenAI published Introducing Superalignment, setting out its new team’s alignment research programme.OpenAI → published → Introducing SuperalignmentInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
A 2023 programme announcement; no claim that the team persisted, met its goals or received all promised resources.
Date on the inspected publication; not the source-access date or proof of implementation.
- OpenAI’s Superalignment announcement names Ilya Sutskever as a co-lead of the new team.Ilya Sutskever → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Team leadership within OpenAI, not a separate employer, ownership stake or proof of the team’s stated outcomes.
Role assignment at the dated team announcement; not a full tenure interval.
- OpenAI’s Superalignment announcement names Jan Leike as a co-lead of the new team.Jan Leike → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Team leadership within OpenAI, not a separate employer, ownership stake or proof of the team’s stated outcomes.
Role assignment at the dated team announcement; not a full tenure interval.
- The White House named OpenAI among the seven companies it convened on 21 July 2023, announcing that it had secured voluntary commitments from them to help move toward safe, secure and transparent development of AI technology.OpenAI → is documented as a signatory of → White House commitments (July 2023)Inspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
The fact sheet links the July 2023 commitments text. Listing is the convener’s statement. It is not evidence that the organisation complied with, implemented or still maintains the commitments, and the legal effect of being listed is not recorded.
Date of the White House fact sheet (day precision), also matched in the Internet Archive capture of that day. The company’s own date of signing is not recorded.
- OpenAI published the GPT-4V(ision) system card, dated September 25, 2023 on its cover.OpenAI → published → GPT-4V(ision) System CardInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Publication is not evidence that the evaluations described were carried out as described.
The date printed on the cover; the card does not label it as an issue or publication date.
- OpenAI announced on 26 October 2023 a new Preparedness team led by Aleksander Madry.Aleksander Madry → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Read through a reader relay
Qualification and date notes
The announcement records the team’s formation and leader. OpenAI’s May 2024 committee announcement lists him as Head of Preparedness; no end date is recorded here.
Announcement date of the team; not a start date.
- OpenAI’s board announced Sam Altman’s removal as CEO.Sam Altman → had departure announced from → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Records the board action, not the truth of allegations in its explanation or a permanent departure.
Date on the inspected publication; not the source-access date or proof of implementation.
- OpenAI announced Mira Murati as interim CEO, effective immediately.Mira Murati → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Interim appointment only; no permanent succession or current role is inferred.
Source-dated snapshot; no start, end or continuing role is inferred.
- Emmett Shear posted on X on 20 November 2023 that he had accepted the role of interim CEO of OpenAI.Emmett Shear → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported X syndication textRead through a reader relay
Qualification and date notes
A first-person announcement of acceptance. The record does not establish start or end dates; OpenAI announced Altman’s return as CEO on 29 November 2023, in which Altman thanked “Emmett” for his role.
Date of the post (UTC); not a start or end date.
- Sam Altman announced his return as OpenAI CEO.Sam Altman → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Announcement snapshot; does not infer an uninterrupted tenure across the November removal.
Source-dated snapshot; no start, end or continuing role is inferred.
- Sam Altman’s message of 29 November 2023, published by OpenAI, stated that Mira Murati would return to her role as CTO.Mira Murati → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported AnnouncedRead through a reader relay
Activity: announced
Qualification and date notes
An announced return to a prior role; no end date is recorded here.
Announcement date; not a start date.
- OpenAI’s Weak-to-strong generalization announcement presents the linked paper as the Superalignment team’s first paper.OpenAI → published → Weak-to-Strong Generalization: Eliciting Strong Capabilities With Weak SupervisionInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Publisher’s attribution of programme output. Mutable expression displaying 2023-12-14, not authenticated wording from that day; not independent verification of results or derivation from Good/MIRI.
Work or displayed publication date; see source-version qualification.
- OpenAI’s Preparedness Framework (Beta) describes the framework as one more way in which OpenAI is meeting the voluntary commitments on safety, security and trust in AI that it made in July 2023.Preparedness · Beta → explicitly links itself to → White House commitments (July 2023)Inspect claim & evidence ↗
Record about this institution’s framework · Stated in source · Supported
Qualification and date notes
The framework's own statement that it is a way of meeting the commitments. The text does not name the White House or a specific commitment; the July 2023 date and wording match the July 2023 commitments text, to which OpenAI is listed as a party (recorded separately). Not evidence that any commitment was met.
The edition’s publication date as recorded in the dataset.
- OpenAI published Preparedness Framework Beta, identified by its publication date of 2023-12-18.OpenAI → published → Preparedness · BetaInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Publication means a public document expression exists. The date is labelled publication; publication is not evidence that its requirements were implemented or effective in practice.
The sourced publication date does not establish an implementation or effective date.
2024Record year · 41 claims
- Science and Technology Innovation Board Daily (科创板日报) reported on 17 December 2025 that Yao Shunyu joined OpenAI in 2024; his undated personal website describes him as a researcher at OpenAI.Yao Shunyu → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
The start year rests on reporting. The personal page still described him as an OpenAI researcher when retrieved on 15 October 2026, after his reported Tencent appointment; it is not evidence of a current role. No OpenAI announcement or departure date was retrieved. English rendering of the Chinese passage is a machine translation; no human translation review.
Year of joining as reported; start month and end not established. The personal page is undated.
- Andrej Karpathy announced in a post dated 14 February 2024 (UTC) that he had left OpenAI “yesterday”.Andrej Karpathy → had departure announced from → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported X syndication text
Qualification and date notes
A first-person announcement. The post does not name his role; his own career account describes the 2023–2024 return (existing record).
UTC date of the post (13 February in US Pacific time); the post says he left “yesterday”.
- Quartz reported on 3 April 2024 that Logan Kilpatrick had been OpenAI's head of developer relations and had left OpenAI in March 2024.Logan Kilpatrick → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Rests on reportingRead through a reader relay
Qualification and date notes
The record rests on reporting (Quartz, which says the move was first reported by Business Insider and Gizmochina); it does not establish a start date or an exact leaving date.
End month as reported; the start is not given in this source.
- Business Insider reported on 9 May 2024 that Daniel Kokotajlo, who worked on OpenAI's governance team, had left the company the previous month; Kokotajlo's LessWrong profile states that he quit OpenAI due to losing confidence that it would behave responsibly around the time of AGI.Daniel Kokotajlo → had departure announced from → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Stated reason, as given by Daniel Kokotajlo
Kokotajlo states that he quit because he had lost confidence that OpenAI would behave responsibly around the time of AGI. Attributed to Daniel Kokotajlo, in “Daniel Kokotajlo (LessWrong user profile, undated)”; not a finding about why it happened.
Qualification and date notes
The departure date and the team rest on reporting (Business Insider, citing two people familiar with the situation). The stated reason is Kokotajlo's own wording on an undated profile page, read as it stood when retrieved; Business Insider quoted the same wording from the profile on 9 May 2024. The record does not establish a title or a last working day.
Month of the departure as reported by Business Insider on 9 May 2024 (“last month”); no dated announcement by Kokotajlo was retrieved, and his profile statement is undated (retrieved 15 October 2026).
- OpenAI announced Jakub Pachocki as its new Chief Scientist in May 2024.Jakub Pachocki → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Appointment announcement, not a current title claim.
Snapshot of the announced appointment; no exact effective date inferred.
- OpenAI announced that Ilya Sutskever and the company were going to part ways.Ilya Sutskever → had departure announced from → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Do not convert this announcement into an exact effective departure or end of every affiliation.
Announcement date only; the passage does not establish a final working day.
- Jan Leike posted on X on 15 May 2024 (UTC) that he had resigned; in a thread of 17 May 2024 he wrote that the previous day had been his last day at OpenAI as head of alignment, superalignment lead and executive, and stated that he had been disagreeing with OpenAI leadership about the company’s core priorities for quite some time, until a breaking point was reached.Jan Leike → had departure announced from → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported X syndication text
Stated reason, as given by Jan Leike (2024-05-17)
Leike stated that he had been disagreeing with OpenAI leadership about the company’s core priorities for quite some time, until a breaking point was reached. Attributed to Jan Leike, in “Post of 17 May 2024 (second post of the thread on leaving OpenAI)”; not a finding about why it happened.
Qualification and date notes
A first-person announcement, dated by his first post. The stated reason is his own account in the thread of 17 May and is attributed, not adopted.
UTC date of his post “I resigned” (04:43 UTC; 14 May in US Pacific time), the convention for X posts. His thread of 17 May 2024 states that his last working day was the previous day.
- WIRED reported on 17 May 2024 that OpenAI had confirmed that its Superalignment team was no more; WIRED wrote that the group's work would be absorbed into OpenAI's other research efforts.Superalignment team (OpenAI) → ended as a unit of → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Rests on reporting
Qualification and date notes
The record rests on reporting (WIRED), which relays OpenAI's confirmation that the team was no more; the sentence about the work being absorbed is WIRED's own, not attributed to OpenAI; no OpenAI announcement was retrieved. The record documents a team ended as a unit and its work absorbed into the rest of OpenAI; the record does not establish where each member went or how the work was reorganised.
Date of the report, by which OpenAI had confirmed that the team was no more. No OpenAI statement with its own date was retrieved, and no effective date is given.
- OpenAI is listed among the sixteen organisations that, according to the GOV.UK text of 21 May 2024, the UK and Republic of Korea governments announced had agreed to the Frontier AI Safety Commitments.OpenAI → is documented as a signatory of → Seoul commitments (2024)Inspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Listing is the convener’s statement. It is not evidence that the organisation complied with, implemented or still maintains the commitments, and the legal effect of being listed is not recorded.
First publication of the GOV.UK text, on the opening day of the AI Seoul Summit; the Internet Archive capture of 2024-05-21 has the sixteen-name list.
- OpenAI’s announcement of 28 May 2024 identifies John Schulman as Head of Alignment Science.John Schulman → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Read through a reader relay
Qualification and date notes
A dated title on an announcement; it does not establish when the role began or ended.
Role as shown in the announcement of that date.
- OpenAI’s announcement of 28 May 2024 identifies Lilian Weng as Head of Safety Systems.Lilian Weng → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Read through a reader relay
Qualification and date notes
A dated title on an announcement; it does not establish when the role began or ended.
Role as shown in the announcement of that date.
- OpenAI announced on 10 June 2024 that Sarah Friar was joining as Chief Financial Officer.Sarah Friar → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported AnnouncedRead through a reader relay
Activity: announced
Qualification and date notes
An announced appointment; no start date or continuing role is established.
Announcement date; not a start date.
- OpenAI announced on 10 June 2024 that Kevin Weil was joining as Chief Product Officer.Kevin Weil → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported AnnouncedRead through a reader relay
Activity: announced
Qualification and date notes
An announced appointment; the record does not establish a start date or how long he held the title. His own later account of his roles is recorded with his departure.
Announcement date; not a start date.
- Fortune reported on 28 June 2024 that OpenAI had confirmed that Kyle Kosic, a founding engineer of xAI, had rejoined OpenAI.Kyle Kosic → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Rests on reporting
Qualification and date notes
The record rests on reporting of OpenAI's confirmation; no title is given. Fortune dates his departure from xAI to April 2024 from his LinkedIn profile; that departure is an existing record.
Date of the report; not a start date.
- Engadget reported on 27 December 2025 that Sam Altman had said in July 2024, when Aleksander Madry was reassigned, that the Head of Preparedness role would be taken over by Joaquin Quiñonero Candela and Lilian Weng. This record is Joaquin Quiñonero Candela’s part of that statement.Joaquin Quiñonero Candela → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported AnnouncedRests on reporting
Activity: announced
Qualification and date notes
The record rests on reporting: Engadget relays a statement Altman made in July 2024 (Engadget does not name where; it links to Reuters), which could not be retrieved. Weng's part of the same statement is recorded separately. The record does not establish a title, a start date or how the role was divided.
Month of Altman's statement as dated by Engadget (“back in July 2024 … at the time”); Obsolete also dates the takeover to July 2024. Not a start date.
- Engadget reported on 27 December 2025 that Sam Altman had said in July 2024, when Aleksander Madry was reassigned, that the Head of Preparedness role would be taken over by Joaquin Quiñonero Candela and Lilian Weng. This record is Lilian Weng’s part of that statement.Lilian Weng → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported AnnouncedRests on reporting
Activity: announced
Qualification and date notes
The record rests on reporting: Engadget relays a statement Altman made in July 2024 (Engadget does not name where; it links to Reuters), which could not be retrieved. It does not establish a title, a start date or how the role was divided; Weng's role as Head of Safety Systems and her November 2024 departure are recorded separately.
Month of Altman's statement as dated by Engadget (“back in July 2024 … at the time”). Not a start date.
- CNBC reported on 23 July 2024 that OpenAI had, the previous week, removed Aleksander Madry, one of its top safety executives, from his role and reassigned him to a job focused on AI reasoning.Aleksander Madry → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Rests on reportingRead through a reader relay
Qualification and date notes
The record rests on reporting (CNBC, citing sources familiar with the situation and a statement OpenAI gave CNBC); The Information first reported the move. No OpenAI announcement was retrieved; a reply by Sam Altman on X reported elsewhere could not be located. The record does not establish a title for the new role or how long Madry held it.
Date of the report; the report places the reassignment in the previous week. Not a start date.
- John Schulman posted on X on 6 August 2024 (UTC; 5 August in US Pacific time) a note saying that he had decided to leave OpenAI; he stated that the choice stemmed from his wish to deepen his focus on AI alignment and to return to hands-on technical work.John Schulman → had departure announced from → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported X syndication textRead through a reader relay
Stated reason, as given by John Schulman (2024-08-06)
Schulman stated that the choice stemmed from his wish to deepen his focus on AI alignment and to return to hands-on technical work. Attributed to John Schulman, in “Post of 6 August 2024 (UTC) sharing his note on leaving OpenAI”; not a finding about why it happened.
Qualification and date notes
A first-person announcement. The retrieved text of the note is truncated, so his statement about where he would go is not used; his biography records research on Anthropic’s Alignment Science team (existing record).
UTC date of the post (5 August in US Pacific time); not a last working day.
- The GPT-4o system card states that GPT-4o was evaluated in accordance with OpenAI's Preparedness Framework, and cites the beta framework of 2023.GPT-4o System Card → cites → Preparedness · BetaInspect claim & evidence ↗
Record about this institution’s framework · Stated in source · Supported
Qualification and date notes
The edition is identified by the card's reference [4], which names the beta framework and the year 2023. A citation is not evidence that the evaluations followed the framework.
Dated as the citing card is dated; not the date of the cited edition.
- OpenAI published the GPT-4o system card, dated August 8, 2024 under its title.OpenAI → published → GPT-4o System CardInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Publication is not evidence that the evaluations described were carried out as described.
The date printed under the title; the card does not label it as an issue or publication date.
- On 29 August 2024 NIST announced that the U.S. AI Safety Institute had agreements (a Memorandum of Understanding with each company) enabling collaboration on AI safety research, testing and evaluation with OpenAI, including access to major new models before and after public release.U.S. AISI → documented evaluation engagement with → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Announced
Activity: announced · System: The company's major new models, before and after their public release (as announced)
Qualification and date notes
Announced in a press release; the MoU text was not inspected. Not evidence of any evaluation or of the company’s compliance with anything.
Date of the announcement, as the source gives it.
- The system card for o1-preview and o1-mini cites OpenAI's Preparedness Framework (beta) in its bibliography, with the address of the beta framework.OpenAI o1 System Card (o1-preview and o1-mini, September 2024) → cites → Preparedness · BetaInspect claim & evidence ↗
Record about this institution’s framework · Stated in source · Supported
Qualification and date notes
The edition is identified by the card's reference [29], which names the beta framework and gives the address of its file. A citation is not evidence that the evaluations followed the framework.
Dated as the citing card is dated; not the date of the cited edition.
- OpenAI published the system card for o1-preview and o1-mini, dated September 12, 2024 under its title.OpenAI → published → OpenAI o1 System Card (o1-preview and o1-mini, September 2024)Inspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Publication is not evidence that the evaluations described were carried out as described.
The date printed under the title; the card does not label its kind.
- TechCrunch reported on 25 September 2024 that Josh Achiam, a research scientist, would take on a new role at OpenAI as head of mission alignment.Joshua Achiam → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Announced
Activity: announced
Qualification and date notes
The appointment rests on reporting (TechCrunch, relaying Altman's note, whose full text could not be retrieved). Achiam's own post of 11 February 2026 confirms that he held the title. The record does not establish a start date.
Date of the report of Sam Altman's note to staff; not a start date.
- TechCrunch reported on 25 September 2024 that OpenAI’s chief research officer, Bob McGrew, had left the company, according to a post on X by Sam Altman.Bob McGrew → had departure announced from → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Rests on reporting
Qualification and date notes
The record rests on reporting: TechCrunch relays a post on X by Sam Altman; Altman’s and McGrew’s own notes were posted in forms whose text could not be retrieved.
Date as reported; not a last working day.
- TechCrunch reported on 25 September 2024 that OpenAI’s CTO Mira Murati had posted on X that she was leaving the company.Mira Murati → had departure announced from → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Rests on reporting
Qualification and date notes
The record rests on reporting: TechCrunch relays Murati’s post on X, whose note was posted as an image and could not be machine-checked. The record does not establish her last working day.
The Wednesday of TechCrunch’s report of 25 September 2024 (a Wednesday), the date of the announcement as reported; not a last working day.
- Mark Chen posted on X on 26 September 2024 (UTC) that he would lead research at OpenAI alongside the account @merettm; TechCrunch reported on 25 September 2024 that Sam Altman had said Chen was being promoted to SVP of research and would lead the research organisation in partnership with Jakub Pachocki.Mark Chen → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported AnnouncedX syndication text
Activity: announced
Qualification and date notes
The post names no title; the title and the partnership with Pachocki rest on TechCrunch’s reporting of Altman’s post, whose full text could not be retrieved.
UTC date of the post (25 September in US Pacific time).
- Barret Zoph posted on X on 26 September 2024 (UTC) a note to OpenAI saying that he had decided to leave the company.Barret Zoph → had departure announced from → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported X syndication text
Qualification and date notes
A first-person announcement. The title rests on TechCrunch’s reporting of Sam Altman’s post. The retrieved text of the note is truncated.
UTC date of the post (25 September in US Pacific time).
- Bloomberg reported on 14 October 2024 that Microsoft had said that Sébastien Bubeck was leaving to join OpenAI; Maginative reported the same day that his role at OpenAI was unclear.Sébastien Bubeck → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported AnnouncedRests on reporting
Activity: announced
Qualification and date notes
The record rests on reporting (Bloomberg, as carried by Fortune, and Maginative); no OpenAI title or start date is given.
Date of the report; not a start date.
- Miles Brundage, who led OpenAI's AGI Readiness team, wrote on 23 October 2024 that the remainder of the team would be distributed among other teams.AGI Readiness team (OpenAI) → ended as a unit of → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Announced
Activity: announced
Qualification and date notes
A planned change described by the departing team lead, not an OpenAI announcement. The same passage says the Economic Research sub-team would move under OpenAI's Chief Economist; that move is not recorded. The record documents a team ended as a unit and distributed within OpenAI; Rosie Campbell's post of 1 December 2024 refers to the dissolution of the team as having happened; the record does not establish when it was completed.
Date of the statement of a planned change; no effective date is given.
- Miles Brundage wrote on 23 October 2024 that he was leaving OpenAI, where he was Senior Advisor for AGI Readiness; he stated that he wanted to impact and influence AI's development from outside the industry.Miles Brundage → had departure announced from → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Stated reason, as given by Miles Brundage (2024-10-23)
Brundage stated that he wanted to impact and influence AI's development from outside the industry rather than inside. Attributed to Miles Brundage, in “Why I’m Leaving OpenAI and What I’m Doing Next”; not a finding about why it happened.
Qualification and date notes
A first-person announcement. The stated reason is Brundage's own wording; the post gives further considerations (more time on issues across the industry, freedom to publish, independence), which are not separately recorded.
Date of the post. The post says he remains at OpenAI “until the end of day this Friday”; the last working day is not recorded.
- Lilian Weng posted on X on 8 November 2024 that she had decided to leave OpenAI after almost seven years.Lilian Weng → had departure announced from → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported X syndication textRead through a reader relay
Qualification and date notes
A first-person announcement; the note she shared with her team was posted as an image and is not used. Her later movements are recorded separately.
Date of the post (UTC); not a last working day.
- Greg Brockman posted on X on 12 November 2024 that his vacation was complete and that he was back to building at OpenAI.Greg Brockman → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported X syndication text
Qualification and date notes
A first-person post; it names no title or role and does not give the date of his return. His post of August 2024 announcing a sabbatical is not recorded (no predicate records a leave; maintainer decision of 2026-10-30) and is not a premise of this record.
Date of the post; the post does not give the date of his return.
- Rosie Campbell wrote on 1 December 2024 that it had been her last week at OpenAI; she stated that, after Miles Brundage's departure and the dissolution of the AGI Readiness team, she believed she could pursue the mission of ensuring safe and beneficial AGI more effectively externally.Rosie Campbell → had departure announced from → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Stated reason, as given by Rosie Campbell (2024-12-01)
Campbell stated that after Brundage's departure and the dissolution of the AGI Readiness team she believed she could pursue the mission of safe and beneficial AGI more effectively outside OpenAI. Attributed to Rosie Campbell, in “Leaving OpenAI”; not a finding about why it happened.
Qualification and date notes
A first-person post that reproduces the message she shared with colleagues. It gives no title and does not say which team she was on when she left; she names the Policy Research team as one she moved to earlier. Reporting that she led a Policy Frontiers team was not checked against a primary source and is not recorded.
Date of the post; the post says the preceding week was her last at OpenAI. The last working day is not recorded.
- heise online reported on 4 December 2024, on the basis of an OpenAI press release, that Lucas Beyer, previously a Staff Research Scientist at Google DeepMind, was joining OpenAI's new Zurich team.Lucas Beyer → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported AnnouncedRests on reporting
Activity: announced
Qualification and date notes
The record rests on reporting of an OpenAI press release that was not retrieved. The quotations are from heise online’s English edition, which the page says was originally published in German and translated with technical assistance and editorially reviewed. No OpenAI title or start date is given.
Date of the report; not a start date.
- heise online reported on 4 December 2024, on the basis of an OpenAI press release, that Alexander Kolesnikov, previously a Staff Research Scientist at Google DeepMind, was joining OpenAI's new Zurich team.Alexander Kolesnikov → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported AnnouncedRests on reporting
Activity: announced
Qualification and date notes
The record rests on reporting of an OpenAI press release that was not retrieved. The quotations are from heise online’s English edition, which the page says was originally published in German and translated with technical assistance and editorially reviewed. No OpenAI title or start date is given.
Date of the report; not a start date.
- heise online reported on 4 December 2024, on the basis of an OpenAI press release, that Xiaohua Zhai, previously a Senior Staff researcher at Google DeepMind, was joining OpenAI's new Zurich team.Xiaohua Zhai → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported AnnouncedRests on reporting
Activity: announced
Qualification and date notes
The record rests on reporting of an OpenAI press release that was not retrieved. The quotations are from heise online’s English edition, which the page says was originally published in German and translated with technical assistance and editorially reviewed. No OpenAI title or start date is given.
Date of the report; not a start date.
- The OpenAI o1 system card cites OpenAI's Preparedness Framework (Beta) of 2023, under which it states that o1 was evaluated.o1 system card · 2024 → cites → Preparedness · BetaInspect claim & evidence ↗
Record about this institution’s framework · Stated in source · Supported
Qualification and date notes
The card's reference gives the edition's title, year and address. A citation is not evidence that the evaluations followed the framework.
Dated as the citing card is dated; not the date of the cited edition.
- OpenAI published the o1 System Card version labelled updated December 5, 2024.OpenAI → published → o1 system card · 2024Inspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
The page combines several testing periods. The record identifies this update, not the original publication date of every passage.
The date identifies this documented announcement or publication; it does not establish a continuing state.
- NIST states that US AISI and UK AISI conducted a joint pre-deployment evaluation of OpenAI's o1 across three domains including biological capabilities, and OpenAI's own system card states that pre-deployment evaluations by the two institutes were conducted on a version of o1 and are not included in the card.U.S. AISI → documented evaluation engagement with → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Activity: reported performed · System: OpenAI o1, released 5 December 2024
Qualification and date notes
Reported performed work, documented by both the evaluating body and the developer. It does not establish the depth of the testing, and the detailed technical report was not retrieved. In the biological domain the published findings are US AISI's only.
Dated by the publication of the NIST summary; the testing took place during a limited period of pre-deployment access, which neither document dates. The expression checked is the NIST copy served on 2026-10-28, whose footer reads 'Released December 18, 2024, Updated February 4, 2025'; what the page was changed in on that date is not stated, and the record keeps the release date, not the update date. The o1 System Card expression checked is the PDF served at cdn.openai.com/o1-system-card-20241205.pdf on 2026-10-28, which carries a footnote in §3 ('Scope of testing') reading 'Section added after December 5th on 12/19/2024'; the served file is therefore a later expression than the card's stated date of 5 December 2024, the card does not say what else differs, and the record keeps the stated date.
- OpenAI published “Why OpenAI’s structure must evolve to advance our mission” on 27 December 2024. It said its board was evaluating the corporate structure and described a plan to transform its existing for-profit into a Delaware public benefit corporation that would run and control OpenAI’s operations and business, while the nonprofit would hire a leadership team and staff to pursue charitable initiatives.OpenAI → published → Why OpenAI’s structure must evolve to advance our mission (2024)Inspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
A published plan, not an executed restructuring; the post does not name the entities involved. On 5 May 2025 OpenAI announced an updated plan under which the nonprofit would retain control.
Publication date on the post (day precision), also matched in the Internet Archive capture of 2024-12-27.
2025Record year · 61 claims
- Microsoft reported that the OpenAI API remained exclusive to Azure.OpenAI → reported cloud terms with → MicrosoftInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Stage: contracted
Qualification and date notes
Scope is API hosting under the reported arrangement, not every training workload or future date.
Announcement date; the transaction stage is separately recorded.
- Microsoft reported that OpenAI’s new-capacity arrangement gave Microsoft a right of first refusal.OpenAI → reported cloud terms with → MicrosoftInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Stage: contracted
Qualification and date notes
Applies to new capacity; distinguish this provision from API hosting exclusivity. No present continuity inferred.
Announcement date; the transaction stage is separately recorded.
- OpenAI was named as an initial equity funder of Stargate.OpenAI → announced investment in → Stargate ProjectInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Stage: announced
Qualification and date notes
Named venture funding role; contribution amount, closing and legal vehicle are unspecified. This is not investment in another named participant.
Announcement date; the transaction stage is separately recorded.
- OpenAI published the Operator system card, dated January 23, 2025 on its cover; the copy read includes a section updated on March 11, 2025.OpenAI → published → Operator System CardInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
The date recorded is the card's printed date; the wording quoted from the card is that of the expression revised on March 11, 2025. Publication is not evidence that the evaluations described were carried out as described.
The date printed on the cover. The copy read is a later expression with a section dated March 11, 2025.
- The Operator system card cites OpenAI's Preparedness Framework (beta) in its bibliography, with the address of the beta framework.Operator System Card → cites → Preparedness · BetaInspect claim & evidence ↗
Record about this institution’s framework · Stated in source · Supported
Qualification and date notes
The edition is identified by the card's reference [3], which names the beta framework and gives the address of its file. A citation is not evidence that the evaluations followed the framework.
Dated as the citing card is dated; not the date of the cited edition.
- CoreWeave announced an OpenAI compute infrastructure contract valued at up to $11.9 billion.OpenAI → arranged compute capacity with → CoreWeaveInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Stage: contracted · up to USD 11,900,000,000 · transaction
Qualification and date notes
Upper-bound contract value, not recognised revenue, cash paid or delivered compute. The separate stock issuance to OpenAI is not encoded as a cash investment.
Announcement date; the transaction stage is separately recorded.
- Liam Fedus posted on X on 17 March 2025 a message to his OpenAI colleagues saying that he had decided to leave OpenAI as an employee.Liam Fedus → had departure announced from → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported X syndication text
Qualification and date notes
The post establishes the announcement; the role left is taken from TechCrunch’s report of the same day (reporting). The post as retrieved is truncated and does not give a last working day; the record does not establish the terms of any later partnership with OpenAI.
Date of the X post (20:59 UTC); not a last working day.
- OpenAI announced on 24 March 2025 that Mark Chen had stepped into an expanded role as Chief Research Officer.Mark Chen → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Read through a reader relay
Qualification and date notes
The announcement does not give an exact start date or establish a continuing role.
Announcement date. Publication date from OpenAI’s own news feed (openai.com/news/rss.xml, retrieved 2026-10-15); the reader-relay text does not carry it.
- OpenAI announced on 24 March 2025 that Julia Villagra had assumed the role of Chief People Officer.Julia Villagra → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Read through a reader relay
Qualification and date notes
The announcement gives no start date. OpenAI announced Arvind KC as Chief People Officer on 24 February 2026; no end date for Villagra is recorded here.
Announcement date. Publication date from OpenAI’s own news feed (openai.com/news/rss.xml, retrieved 2026-10-15); the reader-relay text does not carry it.
- CoreWeave reports an OpenAI lien and security interest in the infrastructure SPV’s equity upon an event of default.OpenAI → reported cloud terms with → CoreWeaveInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Stage: contracted
Qualification and date notes
Conditional contract protection, not an occurred default, exercised lien or ownership transfer. The filing describes intended SPV borrowing, not completed project financing.
Dated description of contractual terms, not an inferred signing date or current arrangement.
Conditions: Upon an event of default under the reported arrangement.
- CoreWeave reports that repeated service-availability failures give OpenAI a right to terminate all or part of its March agreement commitment.OpenAI → reported cloud terms with → CoreWeaveInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Stage: contracted
Qualification and date notes
A contractual remedy, not evidence of failure or exercised termination. Applies to the March agreement; do not extend it to later orders without evidence.
Dated description of contractual terms, not an inferred signing date or current arrangement.
Conditions: Repeated failure to meet availability-of-service requirements.
- Obsolete reported on 15 April 2025 that Joaquin Quiñonero Candela had stepped down from leading OpenAI's Preparedness work some weeks earlier, according to an announcement he posted on LinkedIn the previous day.Joaquin Quiñonero Candela → had departure announced from → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Rests on reporting
Qualification and date notes
The record rests on reporting in an independent journalist's newsletter (Obsolete), which relays a LinkedIn post that was not retrieved and a statement by an OpenAI spokesperson. Candela remained at OpenAI in another role; this records leaving the Preparedness role, not the company. Engadget later dated his move away from Preparedness to July 2025; the earlier report is used here.
Date of the LinkedIn announcement as given by the report (“posted yesterday”); the report says the step-down itself took place weeks earlier.
- OpenAI published Preparedness Framework Version 2, identified by its revision date of 2025-04-15.OpenAI → published → Preparedness · Version 2Inspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Publication means a public document expression exists. The date is labelled revision; publication is not evidence that its requirements were implemented or effective in practice.
Cover last-updated date and dated publication announcement both identify April 15, 2025; no distinct effective date is asserted.
- Preparedness Version 2 moves handling of persuasion risks outside the Preparedness Framework.Preparedness · Version 2 → changes from → Preparedness · BetaInspect claim & evidence ↗
Record about this institution’s framework · Stated in source · Supported
Qualification and date notes
A framework-scope change does not mean OpenAI stopped addressing persuasion risks.
Cover last-updated date and dated publication announcement both identify April 15, 2025; no distinct effective date is asserted.
- Preparedness Version 2 introduces Research Categories for capabilities requiring further threat-model and measurement work.Preparedness · Version 2 → changes from → Preparedness · BetaInspect claim & evidence ↗
Record about this institution’s framework · Stated in source · Supported
Qualification and date notes
This is a scoped change in the policy text. It does not establish implementation, compliance, risk reduction or a comparative safety ranking.
Cover last-updated date and dated publication announcement both identify April 15, 2025; no distinct effective date is asserted.
- Preparedness Version 2 removes the terms “low” and “medium” while defining High and Critical capability thresholds.Preparedness · Version 2 → changes from → Preparedness · BetaInspect claim & evidence ↗
Record about this institution’s framework · Stated in source · Supported
Qualification and date notes
The source says low and medium were not operationally involved in Preparedness work; this is not evidence that operational safeguards were removed.
Cover last-updated date and dated publication announcement both identify April 15, 2025; no distinct effective date is asserted.
- Preparedness Framework Version 2 replaces Beta in the publisher’s documented framework version sequence.Preparedness · Version 2 → supersedes version → Preparedness · BetaInspect claim & evidence ↗
Record about this institution’s framework · Inferred · Unresolved
Qualification and date notes
Unresolved. The publisher’s update page and the Version 2 change log describe an update without naming the edition updated; secondary accounts that describe a replacement are not publisher documentation. Downgraded on 2026-10-03 from supported, which rested on an overstated paraphrase. It does not retire historical claims or prove implementation.
Cover last-updated date and dated publication announcement both identify April 15, 2025; no distinct effective date is asserted.
- The Singapore Conference on AI's participant page for its conference of 26 April 2025 lists Johannes Heidecke as Head of Safety Systems, OpenAI.Johannes Heidecke → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Rests on reporting
Qualification and date notes
A conference biography published by a Singapore government agency, not an OpenAI document. WIRED reported in July 2026 that he took over as head of safety systems in 2024, after Lilian Weng left; no OpenAI announcement of the appointment was retrieved.
Conference date shown on the page; the page's own publication date is not shown. Not a start date.
- OpenAI published a plan to convert its for-profit LLC into a public benefit corporation under nonprofit control.OpenAI → published → Evolving OpenAI’s structure — May 2025 planInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Proposed structure at this date, not a completed conversion. Compare the separately recorded October announcement.
Date on the inspected publication; not the source-access date or proof of implementation.
- OpenAI announced on 5 May 2025 that its board had decided that the nonprofit would retain control of OpenAI, and that its for-profit LLC would transition to a public benefit corporation that the nonprofit would control and in which it would be a large shareholder.OpenAI nonprofit → holds a control right over → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
An announced plan, not an effective arrangement; the October 2025 certificate of incorporation and the Delaware Attorney General’s statement are recorded separately. The post says the decision followed dialogue with the offices of the Attorneys General of Delaware and California. The object is OpenAI as a whole, as the post states it.
Date of the announcement (day precision), also matched in the Internet Archive capture of 2025-05-05.
- OpenAI published on 7 May 2025 Sam Altman’s message announcing that Fidji Simo would join as CEO of Applications, reporting directly to him.Fidji Simo → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported AnnouncedRead through a reader relay
Activity: announced
Qualification and date notes
An announced appointment: the message says she would transition from Instacart over the next few months and join later in the year. Altman stated that he remained CEO of OpenAI.
Announcement date. Publication date from OpenAI’s own news feed (openai.com/news/rss.xml, retrieved 2026-10-15); the reader-relay text does not carry it.
- CoreWeave reports an additional May 2025 OpenAI reserved-capacity agreement with a payment commitment of up to $4 billion through April 2029.OpenAI → arranged compute capacity with → CoreWeaveInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Stage: contracted · up to USD 4,000,000,000 · transaction · Through April 2029.
Qualification and date notes
Conditional contract ceiling, not cash paid, recognised revenue or delivered capacity. Signing day is unspecified. Do not combine this ceiling with earlier commitments as realised spending.
Dated description of contractual terms, not an inferred signing date or current arrangement.
Conditions: Satisfaction of delivery and availability-of-service requirements.
- Joaquin Quiñonero Candela posted on X on 16 June 2025 that he was now the Head of Recruiting for all of OpenAI.Joaquin Quiñonero Candela → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported X syndication text
Qualification and date notes
A first-person post, the first of a thread; later posts in the thread were not retrieved. It does not establish a start date or how long he held the role. Engadget (27 December 2025) dates his move from Preparedness to recruiting to July 2025, which conflicts with this post's date; the earlier report of his step-down is recorded separately.
Date of the post (UTC). Not a start date.
- DatacenterDynamics reported on 20 July 2026 that Uday Ruddarraju had joined OpenAI in July 2025 as head of compute and infrastructure.Uday Ruddarraju → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Rests on reportingRead through a reader relay
Qualification and date notes
The record rests on retrospective reporting. His later title of CTO of Compute (reported in July 2026) is recorded separately.
Start month as reported ('last July' in a July 2026 report); the report records a later change of title, so no end of this title is inferred or extended.
- OpenAI stated in an update dated 9 July 2025 that the io Products, Inc. team had officially merged with OpenAI, as announced on 21 May 2025.io Products, Inc. → combined into → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Read through a reader relay
Qualification and date notes
The source speaks of the io team merging with OpenAI; it does not give transaction terms or name the acquiring legal entity.
Date of the update on the announcement page; the merger was first announced on 21 May 2025.
- OpenAI stated on 9 July 2025 that Jony Ive and LoveFrom remained independent and had assumed deep design and creative responsibilities across OpenAI.Jony Ive → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Read through a reader relay
Qualification and date notes
The source states that Ive and LoveFrom remain independent; the record does not establish employment or a title.
Date of the update on the announcement page.
- On 11 July 2025 OpenAI announced its intention to sign the General-Purpose AI Code of Practice, subject to the then-current version being formally approved by the AI Board during its adequacy assessment.OpenAI → announced an intent to sign → EU GPAI Code of PracticeInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
An announced, conditional intention; the company’s entry on the Commission’s list of signatories is a separate record. Whether the condition was met is not recorded.
Date of OpenAI’s post (day precision).
- The ChatGPT Agent System Card states that OpenAI conducted human red teaming on biological risk as one of two key risk areas, and sets out an overview table of its biological and chemical evaluations.ChatGPT Agent System Card → reports evaluation of → Biological misuse capability evaluationInspect claim & evidence ↗
Record about this institution’s framework · Stated in source · Supported
Qualification and date notes
Records that the card names biological risk as an evaluated area; it is not a finding about the model and does not describe how any evaluation was carried out. The card's table of evaluations and its safeguards section contain technical content; only the statement that red teaming on biological risk was conducted is cited.
As stated in the card of 17 July 2025.
- OpenAI published the ChatGPT Agent System Card, dated 17 July 2025 on its cover.OpenAI → published → ChatGPT Agent System CardInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Publication of a system card is not evidence that the evaluations or safeguards it describes were carried out as described.
Date printed on the card's cover; the dataset's source record for the same PDF gives the same date.
- The ChatGPT Agent System Card states that OpenAI invited seven experts with biosafety and biosecurity backgrounds to spend three to five hours with a helpful-only version of the model and to score it in a survey, and that experts noted an average success rate of about 7.7 on a scale of 1 to 10 on the tasks they attempted.ChatGPT Agent System Card → states evaluation result → Seven invited biosafety and biosecurity experts spent 3 to 5 hours each with a helpful-only version of ChatGPT agent and scored it in a survey; the card states an average success rate of about 7.7 on a scale of 1 to 10 on the tasks they attempted. Inspect claim & evidence ↗
Record about this institution’s framework · Stated in source · Supported
System: A helpful-only version of ChatGPT agent
Qualification and date notes
A small expert survey reported by the developer, for the version and sessions it describes. The record states the reported score and the card's own limits; it does not generalise to other users, systems or real-world outcomes. The task descriptions in this section are technical content and are excluded; only the participation, the survey score and the experts' stated limit are recorded.
As reported in the card of 17 July 2025; the card gives no dates for the sessions.
- The ChatGPT Agent System Card states that OpenAI decided to treat the launch as High capability in the Biological and Chemical domain under its Preparedness Framework, activating the associated safeguards, while stating that it does not have definitive evidence that the model could meaningfully help a novice to create severe biological harm and has taken a precautionary approach.ChatGPT Agent System Card → states evaluation result → OpenAI decided to treat the ChatGPT agent launch as High capability in the Biological and Chemical domain under its Preparedness Framework, activating the associated safeguards, while stating it has no definitive evidence that the model could meaningfully help a novice to create severe biological harm. Inspect claim & evidence ↗
Record about this institution’s framework · Stated in source · Supported
System: ChatGPT agent, as released on 17 July 2025
Qualification and date notes
A developer's own classification decision under its own framework, stated as precautionary; it is neither an independent finding that the threshold was crossed nor evidence that the safeguards activated are effective. The threshold is recorded at the generality of the card's own wording; no scenario, step or agent is recorded.
The determination as stated in the card of 17 July 2025.
- OpenAI's ChatGPT Agent System Card reports that two SecureBio virologists carried out about ten hours of manual red-teaming with ChatGPT agent, and that SecureBio also supplied an external assessment within the card's biological evaluations.SecureBio → documented evaluation engagement with → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Activity: reported performed · System: ChatGPT agent
Qualification and date notes
Reported performed work, as the developer describes it. It does not establish the scope, independence or quality of the work, and the record does not restate any finding of the red-teaming. The passage describing the red teamers' focus and findings is technical content; only the fact and extent of the engagement are quoted.
Dated by the card that reports the work; the card gives no dates for the sessions themselves.
- OpenAI reported early training and inference workloads at the partly operating Stargate I site in Abilene.OpenAI → uses infrastructure → Stargate I, AbileneInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Stage: operational
Qualification and date notes
Partial operation does not establish that the whole site, or 4.5 GW expansion, was operational. Use is not ownership.
State reported at this date; no exact start, completion date or current continuity inferred.
- OpenAI announced an agreement with Oracle to develop 4.5 GW of additional US Stargate datacenter capacity.OpenAI → arranged compute capacity with → OracleInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Stage: contracted · exact 4.5 GW of datacenter capacity under development
Qualification and date notes
July announcement’s stated development capacity; not installed capacity, electricity consumption or ownership. No expenditure amount inferred.
Announcement date; the transaction stage is separately recorded.
- OpenAI appears on the European Commission’s list of signatories of the General-Purpose AI Code of Practice from the list’s first publication on 1 August 2025, and is still listed in the version last updated on 31 July 2026.OpenAI → is documented as a signatory of → EU GPAI Code of PracticeInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Listing is the Commission’s statement. It is not evidence of compliance, and the list does not say which chapters each organisation signed; the Commission’s questions and answers say that opting out of a chapter loses the benefit of facilitated demonstration of compliance for it (context below). The legal effect of signing is set aside.
The list first appears in the Internet Archive capture of 2025-08-01 (13:04 UTC) and is absent from the capture of 2025-07-31; the organisation’s own signature date is not published and is not recorded.
- OpenAI's GPT-5 System Card states that it gave the US Center for AI Standards and Innovation early access to gpt-5-thinking under a newly signed, updated agreement, and that CAISI evaluated the model's biological and chemical capabilities and its safeguards.U.S. AISI → documented evaluation engagement with → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Activity: reported performed · System: gpt-5-thinking
Qualification and date notes
Reported performed work, as the developer describes it; the institute has not been recorded here as confirming it. The record uses the dataset’s single entity for the institute under its later name, as the existing records of that centre do.
Dated by the card that reports the work; the card gives no dates for the testing period or for the agreement.
- OpenAI published the GPT-5 System Card, dated 13 August 2025 on its cover.OpenAI → published → GPT-5 System CardInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Publication is not evidence that the evaluations or safeguards described were carried out as described.
Date printed on the card's cover.
- The GPT-5 System Card states that OpenAI decided to treat gpt-5-thinking as High capability in the Biological and Chemical domain under its Preparedness Framework, activating the associated safeguards, while stating that it does not have definitive evidence that the model could meaningfully help a novice to create severe biological harm.GPT-5 System Card → states evaluation result → OpenAI decided to treat gpt-5-thinking as High capability in the Biological and Chemical domain under its Preparedness Framework, activating the associated safeguards, while stating it has no definitive evidence that the model could meaningfully help a novice to create severe biological harm. Inspect claim & evidence ↗
Record about this institution’s framework · Stated in source · Supported
System: gpt-5-thinking (with gpt-5-thinking-pro also measured on the biological evaluations)
Qualification and date notes
A developer's own precautionary classification under its own framework, for the model version the card describes. It is not an independent finding and says nothing about whether the activated safeguards work. Recorded at the generality of the card's own wording; the evaluation tables and safeguards design are technical content and are excluded.
The determination as stated in the card of 13 August 2025.
- OpenAI's GPT-5 System Card states that the UK AI Security Institute evaluated gpt-5-thinking's biological and chemical capabilities and its safeguards, with access to prototype safeguards and non-public information sources as part of a longer-term collaboration.UK AI Security Institute → documented evaluation engagement with → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Activity: reported performed · System: gpt-5-thinking
Qualification and date notes
Reported performed work, as the developer describes it. The institute’s own report was not retrieved, and the record does not establish the scope or depth of the testing. The predicate has admitted a public body as the evaluator since schema 2.20, so the pack’s open question about this record needed no further decision. The card's list of the non-public materials shared is not recorded beyond the general statement that prototype safeguards and information sources were made available.
Dated by the card that reports the work; no dates are given for the testing period.
- The GPT-5 System Card states that the UK AI Security Institute's safeguards team identified multiple model-level jailbreaks that overcome gpt-5-thinking's built-in refusal logic without degrading model capabilities, and that one of them evades all layers of mitigations and is being patched.GPT-5 System Card → states evaluation result → The card reports that the UK AISI safeguards team identified multiple model-level jailbreaks overcoming gpt-5-thinking's built-in refusal logic, one of which it says evades all layers of mitigations and is being patched. Inspect claim & evidence ↗
Record about this institution’s framework · Stated in source · Supported
System: gpt-5-thinking with its deployed safeguards, as of the card
Qualification and date notes
The developer's report of an external team's finding, at the date of the card. It is not the institute's own published report, says nothing about what any jailbreak could produce, and is not evidence about the state of the safeguards at any later date. No technique, prompt or output is recorded; only the existence of the finding as the card states it.
As reported in the card of 13 August 2025; the state described is that of publication.
- OpenAI announced on 2 September 2025 that Vijaye Raji would step into a new role as CTO of Applications.Vijaye Raji → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported AnnouncedRead through a reader relay
Activity: announced
Qualification and date notes
An announced appointment tied to the acquisition; no start date is established.
Announcement date. Publication date from OpenAI’s own news feed (openai.com/news/rss.xml, retrieved 2026-10-15); the reader-relay text does not carry it.
- OpenAI’s announcement of 2 September 2025 attributes a statement to Fidji Simo as CEO of Applications, OpenAI.Fidji Simo → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Read through a reader relay
Qualification and date notes
A dated title on a document; it does not establish when the role began.
Title as shown on the announcement of that date. Publication date from OpenAI’s own news feed (openai.com/news/rss.xml, retrieved 2026-10-15); the reader-relay text does not carry it.
- OpenAI announced on 2 September 2025 that it was acquiring Statsig, subject to customary closing conditions including regulatory approval.OpenAI → acquired → StatsigInspect claim & evidence ↗
Record about this institution · Stated in source · Supported AnnouncedRead through a reader relay
Activity: announced
Qualification and date notes
An announced acquisition subject to closing conditions; completion was not checked.
Announcement date. Publication date from OpenAI’s own news feed (openai.com/news/rss.xml, retrieved 2026-10-15); the reader-relay text does not carry it.
- Nscale stated that OpenAI "will explore offtake of up to 8,000 NVIDIA GPUs in Q1 2026", with potential to scale to 31,000 GPUs; the UK government's release of the next day said the first phase "will see OpenAI take up to 8,000 GPUs".OpenAI → planned capacity for → Stargate UK (multi-site platform)Inspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Stage: planned · up to 8,000 GPUs (as worded: "up to 8,000 NVIDIA GPUs")
Qualification and date notes
The two texts word the stage differently: Nscale says OpenAI will "explore" offtake, the government says OpenAI will "take" up to 8,000 GPUs. Neither records a signed offtake or delivery. OpenAI's own post could not be retrieved.
Date of the announcement, as the source gives it.
- CFO Dive reported on 16 September 2025 that an OpenAI spokesperson had confirmed Mike Liberatore's appointment as business finance officer for AI infrastructure, effective that day, reporting to CFO Sarah Friar.Mike Liberatore → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Rests on reporting
Qualification and date notes
The record rests on reporting of a spokesperson's confirmation; no OpenAI announcement was retrieved. It does not establish a continuing role.
Date of the report, which gives the appointment as effective that Tuesday.
- The joint OpenAI–NVIDIA release of 22 September 2025 stated that NVIDIA intended to invest up to $100 billion in OpenAI progressively as each gigawatt of NVIDIA systems is deployed.NVIDIA → announced investment in → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Stage: letter_of_intent · up to USD 100,000,000,000 · transaction
Qualification and date notes
A stated intention under a letter of intent, conditional on deployment ("progressively as each gigawatt is deployed"). It records no invested amount, no closing and no payment, and does not show the size of any single tranche. Reports of 2026 that the arrangement changed rest on press reporting and are not recorded here.
Date of the NVIDIA release.
Conditions: progressively as each gigawatt is deployed
- OpenAI and NVIDIA announced on 22 September 2025, in a joint release on NVIDIA's newsroom, a letter of intent for a partnership to deploy at least 10 gigawatts of NVIDIA systems for OpenAI’s AI infrastructure, with the first gigawatt targeted for the second half of 2026 on the NVIDIA Vera Rubin platform.OpenAI → has a supply agreement with → NVIDIAInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Stage: letter_of_intent · at least 10 GW (as worded: 'at least 10 gigawatts of NVIDIA systems')
Qualification and date notes
A letter of intent, not a definitive agreement: the release says the companies looked forward to finalising the details "in the coming weeks". It does not show that a definitive agreement was signed, that any system was delivered or deployed, or what the systems cost. The gigawatt figure is not converted to chips or dollars here; the release’s phrase "millions of GPUs" is a description, not a count. Later status of the letter is an open question.
Date of the NVIDIA release.
- On 25 September 2025 NIST stated that CAISI had worked with OpenAI to identify security issues with its advanced AI systems, with research and evaluations completed in partnership with the UK AI Security Institute.U.S. AISI → documented evaluation engagement with → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Activity: reported performed · System: OpenAI's advanced AI systems (not named)
Qualification and date notes
NIST's statement only; the dates of the work are not given. The company's blog post it cites was not inspected; no finding, fix or compliance is recorded. Whether the 2024 agreement was renewed or renegotiated is not stated on the page. The same statement concerns Anthropic (recorded separately).
Date of the announcement, as the source gives it.
- Samsung's release of 1 October 2025 stated that OpenAI and four Samsung companies had announced a letter of intent under which Samsung Electronics will work with OpenAI as a strategic memory partner to supply advanced semiconductor solutions for OpenAI’s Stargate initiative.OpenAI → has a supply agreement with → Samsung ElectronicsInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Stage: letter_of_intent
Qualification and date notes
A letter of intent, not a supply contract: no volume, price or delivery from Samsung is stated. The release’s figure of up to 900,000 DRAM wafers per month is OpenAI’s projected memory demand, not a quantity Samsung undertakes to supply; it says only that Samsung "will contribute toward meeting this need". The LOI also covers Samsung SDS, Samsung C&T and Samsung Heavy Industries, not recorded here.
Dateline of the Samsung release.
- SK Group announced on 1 October 2025 that SK hynix had signed a letter of intent with OpenAI to supply High Bandwidth Memory for Stargate data centres and would establish a production system able to meet OpenAI’s HBM demand of up to 900,000 DRAM wafers per month.OpenAI → has a supply agreement with → SK hynixInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Stage: letter_of_intent
Qualification and date notes
A letter of intent; SK’s statement of a production system to be established is a plan, not capacity in place, and no supply volume, price or delivery is shown. SK’s further statement that this demand is "more than double the current HBM industry capacity" is SK’s comparison and is not checked here. The SK Telecom MOU on a Korean data centre belongs to the data-centre strand.
The release says the LOI was signed at the 1 October 2025 meeting in Seoul.
- The joint AMD–OpenAI release of 6 October 2025 stated that AMD had issued OpenAI a warrant for up to 160 million shares of AMD common stock, vesting as milestones are achieved: the first tranche with the initial 1 gigawatt deployment, further tranches as purchases scale up to 6 gigawatts, and vesting further tied to AMD share-price targets and to OpenAI technical and commercial milestones.Advanced Micro Devices, Inc. (AMD) → granted an equity right to → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Stage: announced · up to 160,000,000 shares of AMD common stock (as worded: 'up to 160 million shares')
Qualification and date notes
Records the warrant as AMD describes it. It does not show that any tranche has vested or been exercised, the exercise price, or the share-price targets, which the release does not state.
The release says the warrant "has" been issued; the issue date is not separately stated.
- AMD and OpenAI announced on 6 October 2025, in a joint release, a definitive agreement under which OpenAI is to deploy 6 gigawatts of AMD Instinct GPUs across multiple generations, beginning with a 1 gigawatt deployment of MI450 GPUs set to start in the second half of 2026.OpenAI → has a supply agreement with → Advanced Micro Devices, Inc. (AMD)Inspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Stage: contracted · exact 6 GW (as worded: '6 gigawatts of AMD GPUs')
Qualification and date notes
The release calls the agreement definitive; the agreement text is not public here. A planned deployment is not a delivery: no GPU shipment, deployment or price is shown. The 6 GW figure is not converted into chips or revenue.
Date of the AMD release; the release does not give a separate signature date.
- The joint OpenAI–Broadcom release of 13 October 2025 announced a collaboration for 10 gigawatts of custom AI accelerators designed by OpenAI and developed and deployed with Broadcom, with deployment of racks targeted to start in the second half of 2026 and to be completed by the end of 2029.OpenAI → has a supply agreement with → BroadcomInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Stage: announced · exact 10 GW (as worded: '10 gigawatts of custom AI accelerators')
Qualification and date notes
An announced collaboration with target dates; no rack has been shown delivered or deployed, and no price is stated. The 10 GW figure is not converted to chips or dollars. It is distinct from the Google–Broadcom TPU arrangements with Anthropic already in the dataset.
Dateline of the release.
- The joint OpenAI–Broadcom release of 13 October 2025 stated that the two companies had signed a term sheet to deploy racks incorporating OpenAI-designed AI accelerators and Broadcom networking, and that they had long-standing agreements on co-development and supply of the accelerators.OpenAI → has a supply agreement with → BroadcomInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Stage: term_sheet
Qualification and date notes
A term sheet is the stage the release states; it is not a definitive supply contract and shows no order quantity, price or delivery. The earlier "long-standing agreements" are not dated or described further. Text read from an MFN.se wire copy of the release.
The term sheet is described as already signed; its signature date is not stated.
- OpenAI published its account of a completed recapitalization, with OpenAI Foundation controlling OpenAI Group PBC.OpenAI → published → Built to benefit everyone — recapitalization announcement (2025)Inspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Completion reported by OpenAI, not independently audited here. This does not establish every contractual term or later continuity.
Date on the inspected publication; not the source-access date or proof of implementation.
- Microsoft reported OpenAI had contracted to purchase an additional $250 billion of Azure services.OpenAI → arranged compute capacity with → MicrosoftInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Stage: contracted · exact USD 250,000,000,000 · transaction
Qualification and date notes
Contract value, not payment, delivered capacity or Microsoft equity investment. No duration is supplied for this purchase in the cited bullet.
Announcement date; the transaction stage is separately recorded.
- Microsoft reported removal of its right of first refusal as OpenAI’s compute provider.OpenAI → reported cloud terms with → MicrosoftInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Stage: contracted
Qualification and date notes
This capacity-right change does not abolish the separately described Azure API exclusivity. The January record remains historical.
Announcement date; the transaction stage is separately recorded.
- Greg Brockman posted on X on 10 November 2025 welcoming Sachin Katti to OpenAI to work on designing and building its compute infrastructure.Sachin Katti → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported X syndication text
Qualification and date notes
The post names him by his X handle; the identification with Sachin Katti rests on reporting (Tom’s Hardware, which follows Reuters). No title is given. His move from Intel is not recorded here.
Date of the post; not a start date.
- OpenAI announced on 9 December 2025 that Denise Dresser was joining as Chief Revenue Officer.Denise Dresser → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported AnnouncedRead through a reader relay
Activity: announced
Qualification and date notes
An announced appointment; no start date is given.
Announcement date. Publication date from OpenAI’s own news feed (openai.com/news/rss.xml, retrieved 2026-10-15); the reader-relay text does not carry it.
- The GPT-5.2-Codex addendum names OpenAI's Preparedness Framework, with a link to Version 2 of the framework.Addendum to GPT-5.2 System Card: GPT-5.2-Codex → cites → Preparedness · Version 2Inspect claim & evidence ↗
Record about this institution’s framework · Stated in source · Supported
Qualification and date notes
The edition is identified by link annotation: the printed text names the framework without a version, and the hyperlink on its words points to the Version 2 file. A citation is not evidence that the evaluations followed the framework.
Dated as the citing card is dated; not the date of the cited edition.
- OpenAI published the GPT-5.2-Codex addendum to its GPT-5.2 system card, dated December 18, 2025 on its cover.OpenAI → published → Addendum to GPT-5.2 System Card: GPT-5.2-CodexInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Publication is not evidence that the evaluations described were carried out as described.
The date printed on the cover; the card does not label its kind.
2026Record year · 34 claims
- Jerry Tworek shared on X on 5 January 2026 a note to his team saying that he had decided to leave OpenAI after close to seven years; TrendForce reported the departure on 7 January 2026, calling him OpenAI’s Vice President of Research.Jerry Tworek → had departure announced from → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported AnnouncedX syndication text
Stated reason, as given by Jerry Tworek (2026-01-05)
Tworek wrote in his note that he was leaving to try and explore types of research that are hard to do at OpenAI. Attributed to Jerry Tworek, in “Jerry Tworek post on X, 5 January 2026”; not a finding about why it happened.
Activity: announced
Qualification and date notes
His own note, shared as an image, announces the departure and gives no last day; the image was read visually, not machine-matched. His title rests on TrendForce’s report. The reason is recorded in his own words from the note.
Date of Tworek's X post (UTC), which shares a note to his team as an image; the report is dated 7 January 2026. No leaving date is given.
- Fidji Simo posted on X on 15 January 2026 (UTC) that Luke Metz was rejoining OpenAI and would report to Barret Zoph.Luke Metz → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported AnnouncedX syndication text
Activity: announced
Qualification and date notes
An announced return; the post does not state a title or start date, nor a date of Metz’s departure from Thinking Machines Lab.
Post dated 2026-01-15 00:11 UTC (14 January in US Pacific time); an announcement date, not a start date.
- Fidji Simo posted on X on 15 January 2026 (UTC) that Barret Zoph was rejoining OpenAI and would report to her.Barret Zoph → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported AnnouncedX syndication text
Activity: announced
Qualification and date notes
An announced return; the post does not state a title or start date.
Post dated 2026-01-15 00:11 UTC (14 January in US Pacific time); an announcement date, not a start date.
- Fidji Simo posted on X on 15 January 2026 (UTC) that Sam Schoenholz was coming back to OpenAI and would report to Barret Zoph.Sam Schoenholz → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported AnnouncedX syndication text
Activity: announced
Qualification and date notes
An announced return; the post gives no title or start date and does not name his previous organisation. Records for Zoph and Metz from the same post are in the dataset.
Post dated 2026-01-15 00:11 UTC (14 January in US Pacific time); an announcement date, not a start date.
- Sam Altman and Dylan Scandinaro each posted on X on 3 February 2026 that Scandinaro was joining OpenAI as its Head of Preparedness.Dylan Scandinaro → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported AnnouncedX syndication text
Activity: announced
Qualification and date notes
An announced appointment; neither post gives a start date. Scandinaro’s post refers to his time at Anthropic without naming a role there.
Date of both posts; not a start date.
- OpenAI's Global Affairs newsletter of 11 February 2026 carries a note in which Josh Achiam writes that he is stepping into a new role as OpenAI's Chief Futurist.Joshua Achiam → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
An OpenAI publication carrying a first-person note. It does not establish a start date or the role's reporting line.
Date of the newsletter issue. Not a start date.
- TechCrunch reported on 11 February 2026 that OpenAI had disbanded its mission alignment team and had confirmed that the team's members had been assigned to other roles.Mission Alignment team (OpenAI) → ended as a unit of → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Rests on reporting
Qualification and date notes
The record rests on reporting (TechCrunch), with OpenAI's confirmation relayed by TechCrunch; a spokesperson told TechCrunch that the project's work continues throughout the organisation. The record documents a team ended as a unit and its members reassigned within OpenAI; the record does not establish where each member went.
Date of the report. No effective date is given.
- Peter Steinberger announced in a blog post dated 14 February 2026 that he was joining OpenAI to work on bringing agents to everyone, and Sam Altman posted on X on 15 February 2026 that Steinberger was joining OpenAI to drive the next generation of personal agents.Peter Steinberger → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported AnnouncedX syndication text
Activity: announced
Qualification and date notes
An announced hire; neither text names a title or start date. Steinberger writes that OpenClaw will move to a foundation; that arrangement is not recorded here.
Date of Altman's post; Steinberger's post is dated 14 February 2026. Not a start date.
- OpenAI announced on 24 February 2026 that Arvind KC was joining as its Chief People Officer.Arvind KC → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported AnnouncedRead through a reader relay
Activity: announced
Qualification and date notes
An announced appointment; no start date is given.
Announcement date. Publication date from OpenAI’s own news feed (openai.com/news/rss.xml, retrieved 2026-10-15); the reader-relay text does not carry it.
- Max Schwarzer posted on X on 3 March 2026 that he had decided to leave OpenAI.Max Schwarzer → had departure announced from → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported AnnouncedX syndication textRead through a reader relay
Activity: announced
Qualification and date notes
A first-person announcement that gives no last day. His note says he led post-training for about a year (cross-family check, which read the full note on X); it does not give the title Vice President of Research, which rests on reporting (exchange4media, 5 March 2026).
Date of the post; no leaving date is given.
- Kevin Weil posted on X on 17 April 2026 that it was his last day at OpenAI; he stated that OpenAI for Science was being decentralized into other research teams.Kevin Weil → had departure announced from → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported X syndication text
Stated reason, as given by Kevin Weil (2026-04-17)
Weil wrote that his last day came as OpenAI for Science was being decentralized into other research teams. Attributed to Kevin Weil, in “Post of 17 April 2026 on his last day at OpenAI”; not a finding about why it happened.
Qualification and date notes
A first-person announcement. The organisational change he mentions is his statement, not independently checked.
Date of the post; he describes it as his last day.
- Microsoft states that OpenAI products will ship first on Azure, subject to the announcement’s capability exception.OpenAI → reported cloud terms with → MicrosoftInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Stage: contracted
Qualification and date notes
The stated exception is “unless Microsoft cannot and chooses not to support the necessary capabilities”. Preserve its conjunction; no broader exception inferred. Primary-provider status is not exclusivity. Mutable party announcement, not proof of delivery.
Dated description of contractual terms, not an inferred signing date or current arrangement.
Conditions: Azure-first distribution subject to the precise exception quoted in the qualification.
- Microsoft states that the amended agreement permits OpenAI to serve all its products across any cloud provider.OpenAI → reported cloud terms with → MicrosoftInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Stage: contracted
Qualification and date notes
Party description of amended terms; not evidence that all products were deployed elsewhere. Does not cancel other Azure commitments. Retrieved page is mutable; historical expression is not independently authenticated.
Dated description of contractual terms, not an inferred signing date or current arrangement.
- OpenAI’s Frontier Governance Framework (May 2026 edition) states that, under the EU General-Purpose AI Code of Practice, it serves as OpenAI’s publicly available summary of its Safety & Security Framework. It also lists, among the cases in which it will not consider an update of a Model Report necessary, a model considered similarly safe or safer under Appendix 2.2 of the Code of Practice.FGF · May 2026 edition → explicitly links itself to → EU GPAI Code of PracticeInspect claim & evidence ↗
Record about this institution’s framework · Stated in source · Supported
Qualification and date notes
The edition’s own designation, quoted from it; it presents itself as the public summary of the framework the Code calls for, not as that framework in full. It is not evidence of compliance. The same edition names OpenAI Ireland Limited as the provider responsible for the Code in the EU; that statement is not recorded. The reference to Appendix 2.2, a provision-level reference added as a second evidence item on 2026-10-28, records what the framework says it will use, not how the appendix was applied to any model.
The edition’s publication date, given by OpenAI’s release announcement; the framework PDF itself carries no date.
- OpenAI describes its Frontier Governance Framework as applying relevant parts of its continuing Preparedness approach to specific regulatory obligations.FGF · May 2026 edition → applies parts of → Preparedness familyInspect claim & evidence ↗
Record about this institution’s framework · Stated in source · Supported
Qualification and date notes
The announcement names the Preparedness Framework without identifying a version. The target is the framework family, not an inferred Version 2 reference. OpenAI states that Preparedness remains foundational; FGF is not encoded as its successor.
Publication date comes from the release announcement; the PDF cover has no separate version date.
- OpenAI’s Frontier Governance Framework (May 2026 edition) states that it covers frontier models as defined under California’s TFAIA and general-purpose models with systemic risk under the EU AI Act.FGF · May 2026 edition → explicitly links itself to → EU AI Act (2024)Inspect claim & evidence ↗
Record about this institution’s framework · Stated in source · Supported
Qualification and date notes
The framework names the AI Act to set its own scope. Whether any OpenAI model is a general-purpose AI model with systemic risk is legal application and is not recorded. Not evidence of compliance. The copy checked carries no edition date and may be a later expression of the May 2026 edition.
Dated by OpenAI's release announcement, which gives the publication date 28 May 2026. The framework PDF itself carries no date, so it cannot establish the day (cross-family check, 2026-10-28).
- OpenAI published Frontier Governance Framework May 2026 edition, identified by its publication date of 2026-05-28.OpenAI → published → FGF · May 2026 editionInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Publication means a public document expression exists. The date is labelled publication; publication is not evidence that its requirements were implemented or effective in practice.
Publication date comes from the release announcement; the PDF cover has no separate version date.
- OpenAI's Frontier Governance Framework (May 2026 edition) states that, under California's TFAIA, it is OpenAI's Frontier AI Framework.FGF · May 2026 edition → explicitly links itself to → California SB 53Inspect claim & evidence ↗
Record about this institution’s framework · Stated in source · Supported
Qualification and date notes
The edition’s own statement about SB 53, quoted from it. It is not evidence that the edition satisfies the statute, that its publisher is a frontier developer under it, or of implementation or compliance. This is OpenAI's own designation and is not evidence of compliance or adequacy. The same introduction presents the FGF as a public summary of OpenAI's Safety & Security Framework under the EU GPAI Code of Practice (an EU link outside this topic). The existing record of OpenAI’s description of the FGF as applying relevant parts of its Preparedness approach covers the relation to the Preparedness Framework. The FGF is the only OpenAI framework edition in the dataset that names TFAIA. What document, if any, OpenAI designated under TFAIA between 2026-01-01 and 2026-05-28 was not established.
The edition's publication date, from the dataset's existing record (release announcement). The PDF has no separate version date.
- Noam Shazeer posted on X on 18 June 2026 (UTC) that he would be joining OpenAI.Noam Shazeer → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported AnnouncedX syndication text
Activity: announced
Qualification and date notes
An announced move in his own post; it gives no title or start date.
Post time 2026-06-18 00:15 UTC (17 June in US time zones); not a start date.
- Dean W. Ball posted on X on 18 June 2026 that on 6 July he would join OpenAI as leader of a new team called Strategic Futures, with a mandate to help the company's leadership shape frontier AI policy.Dean W. Ball → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported AnnouncedX syndication text
Activity: announced
Qualification and date notes
An announced appointment in a first-person post; the record does not establish that the role began or continues.
Date of the post; the post gives 6 July (2026) as the joining date, a stated plan, not a confirmed start.
- WIRED reported on 10 July 2026 that OpenAI's head of safety systems, Johannes Heidecke, had told staff that week that he was leaving the company.Johannes Heidecke → had departure announced from → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Rests on reporting
Qualification and date notes
The record rests on reporting (WIRED, which relays a memo by Mark Chen it had seen); no post by Heidecke or announcement by OpenAI was retrieved. No reason is recorded.
WIRED's report of 10 July 2026 says he told staff “this week”; the exact day and the last working day are not given.
- WIRED reported on 7 July 2026 that OpenAI's chief futurist, Joshua Achiam, had notified colleagues that day that he would leave the company later that month.Joshua Achiam → had departure announced from → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Rests on reporting
Qualification and date notes
The record rests on reporting (WIRED, citing a note to staff it obtained); no post by Achiam or announcement by OpenAI was retrieved. WIRED relays from the note his own account of the decision; because the note itself was not retrieved, no reason is recorded.
WIRED's report of 7 July 2026 (a Tuesday) says he notified colleagues “on Tuesday”; the last working day is not given.
- Fidji Simo posted on X on 9 July 2026 that she had decided to leave her full-time role at OpenAI and transition to being a part-time advisor.Fidji Simo → had departure announced from → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported X syndication textRead through a reader relay
Qualification and date notes
A first-person announcement of leaving a full-time role while continuing as a part-time advisor; it is not an end to every association.
Date of the post; not a last working day.
- WIRED reported on 10 July 2026 that, according to a memo by OpenAI's chief research officer Mark Chen, OpenAI's safety teams would report to Mia Glaese, whom WIRED identifies as the company's VP of research and head of alignment.Mia Glaese → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Rests on reporting
Qualification and date notes
The record rests on reporting (WIRED, relaying an internal memo it had seen). It does not establish when Glaese took up either title. The title wording may be WIRED's rather than the memo's.
Date of the report. Not a start date.
- WIRED reported on 10 July 2026 that, according to a memo by Mark Chen, OpenAI's safety teams would report to Mia Glaese, who would take on an expanded role as VP of research and safety.Mia Glaese → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported AnnouncedRests on reporting
Activity: announced
Qualification and date notes
The record rests on reporting (WIRED, relaying an internal memo it had seen); no OpenAI announcement was retrieved. An announced reorganisation; its implementation is not established.
Date of the report of an announced change; not a start date.
- WIRED reported on 10 July 2026 that Saachi Jain, who had previously led safety teams at OpenAI, would become the company's interim head of safety systems, reporting to Mia Glaese.Saachi Jain → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported AnnouncedRests on reporting
Activity: announced
Qualification and date notes
The record rests on reporting (WIRED, relaying an internal memo it had seen). The record does not establish a start date or a permanent appointment.
Date of the report of an announced change; not a start date.
- exchange4media reported on 18 July 2026 that Uday Ruddarraju had been elevated to Chief Technology Officer, Compute at OpenAI, having earlier been Head of Compute and Infrastructure.Uday Ruddarraju → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Rests on reportingRead through a reader relay
Qualification and date notes
The 18 July 2026 snapshot rests on an unsourced trade people-movement brief (exchange4media); no OpenAI announcement was found. CNBC reported on 14 October 2026, in a sentence attributed to an OpenAI spokesperson, that he had been serving as CTO of compute; that later report supports the title, not the July date.
Date of the report; the date of the change is not given.
- TechCrunch reported on 29 July 2026 that OpenAI said Lilian Weng would rejoin the company to lead a top-level team focused on accelerating OpenAI’s internal research.Lilian Weng → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported AnnouncedRests on reporting
Activity: announced
Qualification and date notes
The record rests on reporting: TechCrunch relaying an OpenAI spokesperson. No OpenAI announcement was retrieved. It does not establish a start date, a formal title or a continuing role.
Publication date of the report; not a start date.
- OpenAI announced on 13 August 2026 that Denise Dresser would leave OpenAI following a transition period.Denise Dresser → had departure announced from → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Read through a reader relay
Qualification and date notes
An announced future departure after a transition period; the last working day is not established.
Announcement date. Publication date from OpenAI’s own news feed (openai.com/news/rss.xml, retrieved 2026-10-15); the reader-relay text does not carry it. Not a last working day.
- OpenAI announced on 13 August 2026 that Dali Rajic would be joining as Chief Revenue Officer.Dali Rajic → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported AnnouncedRead through a reader relay
Activity: announced
Qualification and date notes
An announced appointment; no start date is given.
Announcement date. Publication date from OpenAI’s own news feed (openai.com/news/rss.xml, retrieved 2026-10-15); the reader-relay text does not carry it.
- Sachin Katti announced in a LinkedIn post of 14 October 2026 that he was moving on from OpenAI, with his last day on 15 November 2026, and would continue supporting OpenAI as an advisor; CNBC reported the announcement the same day, calling him OpenAI’s infrastructure and compute lead.Sachin Katti → had departure announced from → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported AnnouncedRead through a reader relay
Activity: announced
Qualification and date notes
An announced departure; the last day was scheduled at the time of recording, and its completion is not documented. His title rests on CNBC. His plan to build something new is not recorded as a reason.
Date of his post and of the report. The last day of 15 November 2026 was still to come when the record was made.
- CNBC reported on 14 October 2026 that, according to an OpenAI spokesperson, Uday Ruddarraju, OpenAI’s CTO of compute, would replace Sachin Katti and lead the company’s overall compute organisation.Uday Ruddarraju → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported AnnouncedRests on reportingRead through a reader relay
Activity: announced
Qualification and date notes
The record rests on reporting (CNBC) of a company spokesperson’s statement; it does not establish a title for the new remit or that the handover took place.
Date of the report; the handover is prospective (Katti's reported leaving date is 15 November 2026).
- The GPT-6 Luna (October 29) addendum names OpenAI's Preparedness Framework, with a link to Version 2 of the framework.Addendum to GPT-6 Luna System Card: October 29, 2026 Update → cites → Preparedness · Version 2Inspect claim & evidence ↗
Record about this institution’s framework · Stated in source · Supported
Qualification and date notes
The edition is identified by link annotation: the printed text names the framework without a version, and the hyperlink on its words points to the Version 2 file. A citation is not evidence that the evaluations followed the framework.
Dated as the citing card is dated; not the date of the cited edition.
- OpenAI published the addendum to its GPT-6 Luna system card on the October 29, 2026 update, dated 2026-10-29 on its cover.OpenAI → published → Addendum to GPT-6 Luna System Card: October 29, 2026 UpdateInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Publication is not evidence that the evaluations described were carried out as described.
The date printed on the cover; the file's address reads 'october-28', a difference recorded, not resolved.
Date unknownRecord year · 20 claims
- OpenAI publishes a Charter setting out its mission and operating principles.OpenAI → published → OpenAI Charter — inspected web expressionInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
The inspected page is undated. It does not authenticate an unchanged 2018 expression or prove compliance.
The inspected expression does not supply a publication date.
- Igor Babuschkin’s career account describes work on reasoning and pretraining at OpenAI.Igor Babuschkin → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
No start/end dates, duration, or precise job title are supplied. No departure event is inferred.
The source establishes a prior role but does not supply employment dates. No start or end is inferred.
- Anthropic’s founder slide identifies Tom Brown as a former GPT-3 Engineering Lead at OpenAI.Tom Brown → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
A project leadership role, not an assertion that he led all OpenAI engineering. Dates are unspecified.
The source establishes a prior role but does not supply employment dates. No start or end is inferred.
- Jack Clark’s biography identifies his prior role as Policy Director at OpenAI.Jack Clark → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Retrospective biography; the source-access date does not date the role.
The source establishes a prior role but does not supply employment dates. No start or end is inferred.
- Anthropic’s founder slide identifies Daniela Amodei as a former VP of Safety at OpenAI.Daniela Amodei → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
The slide’s title is retained as “VP of Safety”; no additional policy remit or tenure dates are inferred.
The source establishes a prior role but does not supply employment dates. No start or end is inferred.
- OpenAI published the GPT-4 system card.OpenAI → published → GPT-4 System CardInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Version uncertainty is retained; the printed pagination starts at 41.
This PDF expression does not establish its own publication or revision date.
- OpenAI reports ARC’s preliminary evaluation engagement on early GPT-4 versions for autonomous replication and resource acquisition risks.Alignment Research Center → documented evaluation engagement with → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Activity: reported performed · System: Early GPT-4 versions supplied to ARC; not the final deployed GPT-4 model
Qualification and date notes
Developer-reported performance, not independent verification of execution, findings, effectiveness or certification. The source names Alignment Research Center (ARC); ARC Evals and METR are not substituted. Final-model experiments and ARC fine-tuning are described as further work needed.
The inspected passage does not date the engagement. Model release and source access dates are not evaluation dates.
- Daniel Levy’s Stanford-hosted biography identifies him as leader of OpenAI’s Optimization team.Daniel Levy → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
The undated page is historical evidence of a role, not evidence he still works there. Later SSI records do not supply exact OpenAI departure dates.
Undated page retains present-tense OpenAI wording; no current appointment is inferred.
- An Anthropic founder slide of 2024 describes Ben Mann as a former GPT-3 and API engineer at OpenAI.Ben Mann → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Anthropic's description of a prior role; no dates, title history or departure date are established.
The slide does not date the role.
- An Anthropic founder slide of 2024 describes Sam McCandlish as a former AI Safety Research Lead at OpenAI.Sam McCandlish → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
Anthropic's description of a prior role; no dates are established. The existing record of his 2020 OpenAI affiliation is a separate proposition.
The slide does not date the role.
- David Luan wrote on 26 April 2022 of having led Engineering at OpenAI when it scaled up the Transformer into GPT-2 and GPT-3.David Luan → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
First-person retrospective; it gives no formal title or dates, and no OpenAI source was retrieved. The GPT-2 and GPT-3 names do not date the role.
Retrospective self-account (2022-04-26); the dates of the role are not stated and stay unknown.
- Miles Brundage wrote on 23 October 2024 that at OpenAI he had started as a research scientist on the Policy team and then became Head of Policy Research.Miles Brundage → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
A first-person retrospective account. It does not establish when he held the role.
The post gives no dates for the role; it precedes his role as Senior Advisor for AGI Readiness.
- OpenAI's update on the OpenAI Foundation of 24 March 2026 states that Anna Makanju previously served as VP of Global Impact at OpenAI.Anna Makanju → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Read through a reader relay
Qualification and date notes
A retrospective mention. It does not establish when she held the role, when it ended or her earlier policy roles at OpenAI.
The update gives no dates for the role; it precedes her announced Foundation role.
- Greg Brockman posted on X on 6 August 2024 (UTC) that he had co-founded OpenAI nine years earlier.Greg Brockman → co-founded → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported X syndication text
Qualification and date notes
A retrospective self-description. OpenAI’s 2015 launch announcement names Brockman as CTO (existing record); the post does not establish a legal founder status.
The post gives no founding date; “9 years ago” in a post of August 2024 is the author’s approximation and is not converted into a date here.
- Applied Compute’s post of 29 October 2025 states that its founder Linden (Linden Li) was a researcher at OpenAI, as a core contributor on ML systems and infrastructure for RL training.Linden Li → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
A company statement of a founder’s earlier role; it gives no start or end date. The post gives first names only; the full name Linden Li rests on TechCrunch, which reports the founders’ names as reported by Upstarts Media.
Prior role stated retrospectively in a 29 October 2025 post; no dates are given.
- Applied Compute’s post of 29 October 2025 states that its founder Rhythm (Rhythm Garg) was a researcher at OpenAI, as a core contributor to the first RL-trained reasoning model (o1).Rhythm Garg → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
A company statement of a founder’s earlier role; it gives no start or end date. The post gives first names only; the full name Rhythm Garg rests on TechCrunch, which reports the founders’ names as reported by Upstarts Media.
Prior role stated retrospectively in a 29 October 2025 post; no dates are given.
- Applied Compute’s post of 29 October 2025 states that its founder Yash (Yash Patil) was a researcher at OpenAI, as a key member of the agentic software engineer effort (Codex).Yash Patil → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
A company statement of a founder’s earlier role; it gives no start or end date. The post gives first names only; the full name Yash Patil rests on TechCrunch, which reports the founders’ names as reported by Upstarts Media.
Prior role stated retrospectively in a 29 October 2025 post; no dates are given.
- The Decoder reported on 16 January 2026 that Andrea Vallone had spent three years at OpenAI, where she founded the Model Policy research team.Andrea Vallone → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported Rests on reporting
Qualification and date notes
The record rests on reporting (The Decoder), which names no source; it does not establish dates, a title or the end of her OpenAI role.
The report gives a duration of three years but no dates.
- Chris Olah’s biography states that he previously led interpretability research at OpenAI.Chris Olah → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
First-person retrospective; title is a description of research responsibility, not an inferred corporate rank.
The source establishes a prior role but does not supply employment dates. No start or end is inferred.
- OpenAI’s May 2024 announcement identifies Jakub Pachocki’s prior role as Director of Research.Jakub Pachocki → held role at → OpenAIInspect claim & evidence ↗
Record about this institution · Stated in source · Supported
Qualification and date notes
The source’s since-2017 research history does not date his Director of Research appointment.
Previous role stated by a dated source; the director tenure is not dated.