{"entities":[{"id":"openai","name":"OpenAI"},{"id":"xai","name":"xAI"},{"id":"eliezer-yudkowsky","name":"Eliezer Yudkowsky"},{"id":"greg-yang","name":"Greg Yang"},{"id":"rsp-v2-1","name":"Anthropic Responsible Scaling Policy — v2.1"},{"id":"rsp-v3-1","name":"Anthropic Responsible Scaling Policy — v3.1"},{"id":"rsp-v3-2","name":"Anthropic Responsible Scaling Policy — v3.2"},{"id":"fsf-v3-1","name":"Google DeepMind Frontier Safety Framework — v3.1"},{"id":"faif-2026","name":"xAI Frontier Artificial Intelligence Framework — June 2026 edition"},{"id":"oracle","name":"Oracle"},{"id":"mistral-ai","name":"Mistral AI"},{"id":"drexler-1986-engines-of-creation","name":"Engines of Creation: The Coming Era of Nanotechnology"},{"id":"extropians-1996-09-04-bostrom","name":"Re: Thinking about the future..."},{"id":"recursive-self-improvement","name":"Recursive self-improvement"},{"id":"ai-automated-ai-research","name":"AI-automated AI research"},{"id":"default-outcome-good","name":"It will probably turn out fine"},{"id":"sl4-2000-12-12-yudkowsky-0012-0381","name":"Re: Revising a Friendly AI"},{"id":"sl4-2001-02-07-yudkowsky-0102-0545","name":"Re: Learning to be evil"},{"id":"european-parliament-and-council","name":"European Parliament and Council of the European Union"},{"id":"eu-digital-omnibus-ai-2026","name":"Regulation (EU) 2026/1744 (Digital Omnibus on AI), as published in the Official Journal"},{"id":"unautomated-tasks-bottleneck-growth","name":"Essential unautomated tasks bottleneck growth"},{"id":"other-constraints-prevent-explosive-growth","name":"Other constraints prevent explosive growth"},{"id":"hanson-economic-growth-given-machine-intelligence","name":"Economic Growth Given Machine Intelligence"},{"id":"erdil-besiroglu-2023-explosive-growth-review","name":"Explosive growth from AI automation: A review of the arguments"},{"id":"mansfield-1966-technological-change","name":"Technological Change: Measurement, Determinants, and Diffusion"},{"id":"ai-improving-ai-accelerates-capability","name":"Machine intelligence improving itself accelerates its capability growth"},{"id":"hanson-2008-ai-go-foom","name":"AI Go Foom"},{"id":"adept","name":"Adept"},{"id":"meta-superintelligence-labs","name":"Meta Superintelligence Labs"},{"id":"trapit-bansal","name":"Trapit Bansal"},{"id":"zach-brock","name":"Zach Brock"},{"id":"edwin-mansfield","name":"Edwin Mansfield"},{"id":"g7-hiroshima-code-of-conduct-2023","name":"Hiroshima Process International Code of Conduct for Organizations Developing Advanced AI Systems"},{"id":"cn-mofcom-2025-61","name":"MOFCOM Announcement No. 61 of 2025 on export controls on related foreign rare-earth items (商务部公告2025年第61号)"},{"id":"cn-mofcom","name":"Ministry of Commerce of the People’s Republic of China (中华人民共和国商务部)"},{"id":"batalis-ai-biorisk-explainer-2023-12","name":"AI and Biorisk: An Explainer"},{"id":"ab-prop-physical-bottleneck","name":"The binding barrier is physical execution rather than information"}],"claims":[{"id":"rsp-v3-2-reviewer-approval","statement":"RSP v3.2 gives the Long-Term Benefit Trust authority to approve the selection of external Risk Report reviewers.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"This is a scoped change in the policy text. It does not establish implementation, compliance, risk reduction or a comparative safety ranking.","evidenceIds":["evidence-rsp-v3-2-reviewer-approval"],"attributes":{"changeBasis":"changelog"}},{"id":"openai-oracle-capacity-2025","statement":"OpenAI announced an agreement with Oracle to develop 4.5 GW of additional US Stargate datacenter capacity.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"July announcement’s stated development capacity; not installed capacity, electricity consumption or ownership. No expenditure amount inferred.","evidenceIds":["evidence-openai-oracle-capacity-2025"],"attributes":{"transactionType":"compute_capacity","relationshipContext":"external","transactionStage":"contracted","quantity":{"value":4.5,"unit":"GW of datacenter capacity under development","bound":"exact"}}},{"id":"prov-rsp-v2-1-misuse-uplift-threshold","statement":"RSP v2.1 states that its CBRN-3 Capability Threshold is the ability to significantly help individuals or groups with basic technical backgrounds (e.g., undergraduate STEM degrees) create/obtain and deploy CBRN weapons.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"What the edition states, quoted from the cited expression. It is not evidence of implementation, compliance or effectiveness.","evidenceIds":["evidence-prov-rsp-v2-1-misuse-uplift-threshold"],"attributes":{"provision":{"kind":"threshold","key":"rsp-misuse-uplift-threshold"}}},{"id":"ts-a5-drexler-1986","statement":"Engines of Creation expects automated engineering systems to pull ahead on their own and to help develop the molecular technology that makes faster AI systems.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"A forecast; the wording is attested from 1997.","evidenceIds":["evidence-ts-a5-drexler-1986"],"attributes":{"thought":{"roles":["articulation"],"register":"nonfiction","attested":{"date":"1997-07-09","note":"Wording checked in the Foresight Institute’s web edition (© 1986 K. Eric Drexler; published for the web by Russell Whitaker), whose earliest captures are of 1997-07-09; a reader’s message of October 1996 on the site shows the edition was online by then. Not compared with the 1986 printing."}}}},{"id":"ts-a3-drexler-1986-context","statement":"Engines of Creation has automated engineering help build faster, assembler-built AI systems.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Successor construction through manufacturing, not self-improvement under the criterion.","evidenceIds":["evidence-ts-a3-drexler-1986-context"],"attributes":{"thought":{"roles":["context"],"register":"nonfiction","missing":"The faster engineering comes from new, assembler-built AI systems; the passage does not have one system improve its own capacity to improve.","attested":{"date":"1997-07-09","note":"Wording checked in the Foresight Institute’s web edition (© 1986 K. Eric Drexler; published for the web by Russell Whitaker), whose earliest captures are of 1997-07-09; a reader’s message of October 1996 on the site shows the edition was online by then. Not compared with the 1986 printing."}}}},{"id":"ts-a3-bostrom-1996-09-04","statement":"Bostrom’s message of 1996-09-04 supposes that an AI, once sufficiently intelligent, could think out a radical improvement to its own design, which would make it more intelligent and let it accelerate further. Recorded as articulating the proposition “Recursive self-improvement”.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Bostrom raises this to challenge Hanson and adds \"(I think this is Dan Clemmensen’s view.)\"; the attribution is the message’s own, not evidence of influence.","evidenceIds":["evidence-ts-a3-bostrom-1996-09-04"],"attributes":{"thought":{"roles":["articulation"],"register":"nonfiction","qualifier":"Presented as a supposition and attributed to Dan Clemmensen."}}},{"id":"yk-y30-sl4-2001-02-07-yudkowsky-0102-0545-c","statement":"The SL4 message “Re: Learning to be evil” (2001-02-07) contests Y30. Replying to Gordon Worley on whether AIs learn to be evil, he says AIs have neither evil impulses nor goodness by default unless designed in.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Premise addressed: The default outcome is good. 'em' refers to evil impulses. Framed as an engineering-conservative assumption, not a probability estimate; in the same message he calls convergent altruism a 'pleasant and significant possibility' he is not relying on. \"'em\" means \"evil impulses\". Answering the claim that Friendliness is inherent in superintelligences.","evidenceIds":["evidence-yk-t30-s2"],"attributes":{"thought":{"roles":["contesting"],"register":"nonfiction","engagement":{"kind":"engages-premise","rationale":"Under the message-level rule, the passage, read with the paragraph, message or dialogue turn it belongs to, gives a reason against the premise that the default outcome is good."}}}},{"id":"omni-enacted","statement":"The European Parliament and the Council adopted Regulation (EU) 2026/1744, the Digital Omnibus on AI, which amends Regulation (EU) 2024/1689 and is dated 8 July 2026; it was published in the Official Journal on 24 July 2026.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Adoption establishes the text and its date. It does not show how any amended provision is applied. This record covers only the amendments that bear on general-purpose AI model providers and the AI Act’s dates; the Regulation also amends Regulations (EU) 2018/1139 and (EU) 2023/1230 and many provisions on high-risk AI systems, which are not recorded.","evidenceIds":["evidence-omni-enacted-1","evidence-omni-enacted-2","evidence-omni-enacted-3","evidence-omni-enacted-4"]},{"id":"ec-g04-hanson-economic-growth-given-machine-intelligence-c","statement":"Hanson, “Economic Growth Given Machine Intelligence” (2000-08-19) contests G04. In the paper's example, a residue of human-only jobs lengthens the economic doubling time from 18 to 27 months, against 16 years without machine intelligence: growth stays far above historical rates.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Reclassified from bears-on in the correction pass (audits B and C).","evidenceIds":["evidence-ec-g04-e"],"attributes":{"thought":{"roles":["earliest-located","contesting"],"register":"nonfiction","claimed":{"date":"1998","note":"Author's own citation in the December 2000 paper. The PDF gives the author's Berkeley address; his vita lists a Berkeley postdoc in 1997–1999 and a George Mason post from 1999. The 1998 wording was not retrieved."},"attested":{"date":"2000-08-19","note":"Capture of the PDF (PDF metadata: created 2000-06-22). No earlier capture was retrieved: year-only Wayback requests for this URL and for hanson.berkeley.edu/aigrow.pdf found none, and the CDX index was unavailable during retrieval, so an earlier capture is not excluded."},"engagement":{"kind":"contrary-only","rationale":"A model result under fixed product shares (Cobb-Douglas form); it does not argue against the premise that unautomated tasks are strong complements."},"tier":{"value":1,"basis":"Tier 1 (unchanged). Working paper with a formal growth model, abstract and references; the checked file gives an econ.berkeley.edu contact address, and the author's December 2000 working paper is issued from the Department of Economics, George Mason University."},"conditions":"Product shares fixed; human-only jobs keep a share of 0.05 of output in the example."}}},{"id":"ec-g05-erdil-besiroglu-2023-explosive-growth-review-c","statement":"Erdil and Besiroglu, “Explosive growth from AI automation: A review of the arguments” (2023-10-01) contests G05. Accepts eventual limits and contests their timing, with an estimate of returns to scale.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The record documents what the text says; it does not establish that the argument or model is right.","evidenceIds":["evidence-ec-g05-g"],"attributes":{"thought":{"roles":["contesting"],"register":"nonfiction","claimed":{"date":"2023-09-20","note":"Date of v1, which was not retrieved."},"attested":{"date":"2023-10-01","note":"Version 2 (2023-10-01) is the earliest retrieved version; v1 has no capture. The v2 PDF itself was captured in 2025."},"engagement":{"kind":"engages-premise","rationale":"Disputes the premise that resource constraints bind soon enough to preclude explosive growth."},"tier":{"value":1,"basis":"Tier 1 (re-tiered on 2026-10-25). A scholarly review with an economist author: Tamay Besiroglu held an MPhil in Economics from the University of Cambridge before the text, as his own site says in a capture of 2021-09-15 and his MIT FutureTech profile states (degree of 2020); MIT FutureTech is his affiliation on the v2 title page. This reverses the tier-2 borderline settled on 2026-10-11, whose premise, that neither author was shown to be an economist, did not hold."}}}},{"id":"ec-e01-hanson-2008-ai-go-foom-a","statement":"Hanson, “AI Go Foom” (2008-11-20), in “The Hanson-Yudkowsky AI-Foom Debate”, articulates E01. Hanson summarises Yudkowsky's argument \"as best I can, so I can then respond\".","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The passage reports a view the author does not state as their own. Hanson's paraphrase of Yudkowsky's argument, not his own view: he summarises it \"as best I can, so I can then respond\". Tier 2. Post date 20 November 2008, from the original Overcoming Bias page (datePublished 2008-11-20T03:00:00Z, retrieved 2026-10-15). The compilation's heading \"10 November 2008\" is wrong: the post quotes \"Setting the Stage\" (18 November 2008), and the compilation places it between posts of 20 and 21 November. The UTC timestamp may correspond to the evening of 19 November in US time; the 2008 page was not retrieved.","evidenceIds":["evidence-ec-e01-08"],"attributes":{"thought":{"roles":["articulation"],"register":"relay","tier":{"value":2,"basis":"Blog debate (2008 Overcoming Bias posts) between an economist and a non-economist: tier 2 as an online debate; the maintainer confirmed tier 2 for the debate on 2026-10-12."}}}},{"id":"lg-new-mistral-founding-2023","statement":"Mistral AI’s company page states that Mistral was born in April 2023.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"A company’s own retrospective account; it does not establish the date of legal incorporation.","evidenceIds":["lg-ev-new-mistral-founding-2023"]},{"id":"lg-deal-brock-adept-ceo-2024","statement":"Adept announced on 28 June 2024 that Zach Brock, its Head of Engineering, would be taking over as CEO.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"An announced appointment; the record does not establish a start date or a continuing role.","evidenceIds":["lg-ev-deal-brock-adept-ceo-2024"],"attributes":{"activityStage":"announced","role":"CEO"}},{"id":"lg-mx-bansal-msl-2025","statement":"Mark Zuckerberg's memo of 30 June 2025, as published by CNBC, names Trapit Bansal, described as a co-creator of OpenAI's o-series models, among new MSL team members.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The record rests on reporting: the memo text as published by CNBC. Prior work is Meta's characterisation.","evidenceIds":["lg-ev-mx-bansal-msl-2025"],"attributes":{"role":"Team member, Meta Superintelligence Labs"}},{"id":"lg-mx-yang-cofounded-xai","statement":"Business Insider reported on 25 March 2026 that Greg Yang was among xAI's cofounders.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The record rests on reporting; Yang is named in a list of cofounder departures. Not a legal founder record.","evidenceIds":["lg-ev-mx-yang-cofounded-xai"],"attributes":{"role":"Cofounder"}},{"id":"ap-edwin-mansfield-mansfield-1966-technological-change","statement":"“Technological Change: Measurement, Determinants, and Diffusion” (1966-02) credits Edwin Mansfield as its author.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Credited authorship of this expression; it says nothing about priority for the ideas discussed.","evidenceIds":["evidence-ap-edwin-mansfield-mansfield-1966-technological-change"]},{"id":"ap-eliezer-yudkowsky-sl4-2000-12-12-yudkowsky-0012-0381","statement":"The mailing-list message “Re: Revising a Friendly AI” (2000-12-12) is sent under the name Eliezer S. Yudkowsky.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"A sender name on a mailing-list message as archived; it does not establish identity beyond the credited name, and quoted text by others inside the message is not the sender’s.","evidenceIds":["evidence-ap-eliezer-yudkowsky-sl4-2000-12-12-yudkowsky-0012-0381"]},{"id":"cmp-fsf-v3-1-d5-cadence","statement":"FSF · v3.1 states a provision on the comparison question “Evaluation timing”.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Material capability change assessments are run on various checkpoints once a post-training run is completed (PDF p. 5). For later versions deployed externally, Google DeepMind decides if another critical capability assessment is needed where a model shows meaningful new capabilities or a material increase in performance and it believes these could materially undercut its justification for accepting the risks. No compute or calendar interval is stated. Footnote 4 (PDF p. 5) says the assessment may be skipped for low-risk external deployments, such as to a few trusted testers, if the governance function finds the residual risk acceptable, even where a T/CCL has been reached. What the edition states, paraphrased from the cited expression with its page. It is not evidence of implementation, compliance or effectiveness, and a shared question does not make two editions’ provisions equivalent.","evidenceIds":["ev-cmp-fsf-v3-1-d5-cadence-1"],"attributes":{"provision":{"kind":"evaluation_timing","key":"comparison-d5-cadence","facet":{"id":"D5.cadence","status":"stated"}}}},{"id":"cmp-faif-2026-d7-internal","statement":"FAIF · June 2026 edition states a provision on the comparison question “Internal reporting”.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"No internal recipient is named. Employee escalation is listed as a channel for identifying incidents (PDF p. 8). What the edition states, paraphrased from the cited expression with its page. It is not evidence of implementation, compliance or effectiveness, and a shared question does not make two editions’ provisions equivalent.","evidenceIds":["ev-cmp-faif-2026-d7-internal-1"],"attributes":{"provision":{"kind":"reporting","key":"comparison-d7-internal","facet":{"id":"D7.internal","status":"stated"}}}},{"id":"fl-intl-g7-code-testing","statement":"Action 1 of the Hiroshima Process Code of Conduct calls for appropriate measures throughout development, including before and during deployment, to identify, evaluate and mitigate risks, including diverse internal and independent external testing such as red-teaming.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"A voluntary action; the code says organisations should follow the actions in line with a risk-based approach.","evidenceIds":["evidence-fl-intl-g7-code-testing-1"],"attributes":{"provision":{"kind":"required_action","key":"intl-g7-code-testing"}}},{"id":"cpo-ch-cn-mofcom-2025-61-issued","statement":"MOFCOM issued Announcement No. 61 of 2025 on 9 October 2025, requiring foreign exporters to obtain a MOFCOM dual-use export licence for certain foreign-made items containing Chinese-origin rare-earth items, with parts effective from 1 December 2025 and the rest on publication.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Records issuance and the licensing requirement in summary; the item list (attachment) was not retrieved. The announcement was suspended on 7 November 2025 (recorded separately).","evidenceIds":["evidence-cpo-ch-cn-mofcom-2025-61-issued-1","evidence-cpo-ch-cn-mofcom-2025-61-issued-2"]},{"id":"ab-arg-d2-batalis-context","statement":"CSET's explainer \"AI and Biorisk\" (December 2023) states that the barrier to scientific information is not high even for beginners, and that feasibility depends on both information and hands-on production success, without stating which of the two limits biological misuse.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The record documents what this explainer states; it is not evidence that the proposition is true and implies no influence between texts. The explainer bears on the proposition without meeting its criterion, so the record is not an articulation and holds no earliest-located role. The explainer describes laboratory steps in the surrounding text. Only its conclusion sentences are cited; that content was excluded and is not paraphrased.","evidenceIds":["evidence-ab-arg-d2-batalis-context-1","evidence-ab-arg-d2-batalis-context-2","evidence-ab-arg-d2-batalis-context-3"],"attributes":{"thought":{"roles":["context"],"register":"nonfiction","qualifier":"Stated across two of the explainer's own conclusions — that the information barrier is not high and that feasibility also requires hands-on production success — and in no passage that ranks the two.","missing":"The explainer states that information and hands-on production success are both required; it does not state that access to information is not the barrier that limits biological misuse."}}}],"sources":[{"id":"source-rsp-2-1","title":"Responsible Scaling Policy — v2.1","authors":["Anthropic"],"publisher":"Anthropic","publicationDate":null,"version":"Responsible Scaling Policy v2.1; effective date 2025-03-31.","versionDate":"2025-03-31","canonicalUrl":"https://www-cdn.anthropic.com/17310f6d70ae5627f55313ed067afc1a762a4068.pdf","evidenceUrl":"https://www-cdn.anthropic.com/17310f6d70ae5627f55313ed067afc1a762a4068.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20250402134435/https://www-cdn.anthropic.com/17310f6d70ae5627f55313ed067afc1a762a4068.pdf","archivedAt":"2025-04-02","accessedAt":"2026-09-18","reuse":{"attribution":"Anthropic, Responsible Scaling Policy — v2.1.","licence":"No open redistribution licence asserted","note":"Selekoa practice: metadata, attributed paraphrases and short attributed quotations under the editorial policy (at most 50 words each; at most 400 quoted words per document across the dataset, and about 10% of a document under about 2,000 words). Documents are linked externally, not redistributed. Publisher terms that forbid quotation override this practice."}},{"id":"source-rsp-3-2","title":"Responsible Scaling Policy — v3.2","authors":["Anthropic"],"publisher":"Anthropic","publicationDate":null,"version":"Responsible Scaling Policy v3.2; effective date 2026-04-29.","versionDate":"2026-04-29","canonicalUrl":"https://cdn.sanity.io/files/4zrzovbb/website/28c6241900d90410628a8a2003a5572faae4365a.pdf","evidenceUrl":"https://cdn.sanity.io/files/4zrzovbb/website/28c6241900d90410628a8a2003a5572faae4365a.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20260505141510/https://cdn.sanity.io/files/4zrzovbb/website/28c6241900d90410628a8a2003a5572faae4365a.pdf","archivedAt":"2026-05-05","accessedAt":"2026-09-18","reuse":{"attribution":"Anthropic, Responsible Scaling Policy — v3.2.","licence":"No open redistribution licence asserted","note":"Selekoa practice: metadata, attributed paraphrases and short attributed quotations under the editorial policy (at most 50 words each; at most 400 quoted words per document across the dataset, and about 10% of a document under about 2,000 words). Documents are linked externally, not redistributed. Publisher terms that forbid quotation override this practice."}},{"id":"source-fsf-3-1","title":"Frontier Safety Framework — v3.1","authors":["Google DeepMind"],"publisher":"Google DeepMind","publicationDate":"2026-04-17","version":"Frontier Safety Framework v3.1; publication date 2026-04-17.","versionDate":"2026-04-17","canonicalUrl":"https://storage.googleapis.com/deepmind-media/DeepMind.com/Blog/strengthening-our-frontier-safety-framework/frontier-safety-framework_3-1.pdf","evidenceUrl":"https://storage.googleapis.com/deepmind-media/DeepMind.com/Blog/strengthening-our-frontier-safety-framework/frontier-safety-framework_3-1.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20260423021420/https://storage.googleapis.com/deepmind-media/DeepMind.com/Blog/strengthening-our-frontier-safety-framework/frontier-safety-framework_3-1.pdf","archivedAt":"2026-04-23","accessedAt":"2026-09-18","reuse":{"attribution":"Google DeepMind, Frontier Safety Framework — v3.1.","licence":"No open redistribution licence asserted","note":"Selekoa practice: metadata, attributed paraphrases and short attributed quotations under the editorial policy (at most 50 words each; at most 400 quoted words per document across the dataset, and about 10% of a document under about 2,000 words). Documents are linked externally, not redistributed. Publisher terms that forbid quotation override this practice."}},{"id":"source-faif-2026","title":"Frontier Artificial Intelligence Framework — June 2026 edition","authors":["xAI"],"publisher":"xAI","publicationDate":null,"version":"Frontier Artificial Intelligence Framework June 2026 edition; effective date 2026-06-30. Visible public cover identifies the effective edition; stale embedded PDF draft metadata is not used to infer publication status.","versionDate":"2026-06-30","canonicalUrl":"https://media.x.ai/v1/website/xai-frontier-artificial-intelligence-framework-30-june-2026-99c40684.pdf","evidenceUrl":"https://media.x.ai/v1/website/xai-frontier-artificial-intelligence-framework-30-june-2026-99c40684.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20260716003429/https://media.x.ai/v1/website/xai-frontier-artificial-intelligence-framework-30-june-2026-99c40684.pdf","archivedAt":"2026-07-16","accessedAt":"2026-09-18","reuse":{"attribution":"xAI, Frontier Artificial Intelligence Framework — June 2026 edition.","licence":"No open redistribution licence asserted","note":"Selekoa practice: metadata, attributed paraphrases and short attributed quotations under the editorial policy (at most 50 words each; at most 400 quoted words per document across the dataset, and about 10% of a document under about 2,000 words). Documents are linked externally, not redistributed. Publisher terms that forbid quotation override this practice."}},{"id":"source-oracle-stargate-2025","title":"Stargate advances with 4.5 GW partnership with Oracle","authors":["OpenAI"],"publisher":"OpenAI","publicationDate":"2025-07-22","version":"Dated announcement inspected 2026-09-19; a historical company statement, not independent verification of implementation. Archived copy: the Wayback Machine capture of 2025-07-22 11:51:19 UTC, the earliest same-day capture found by a bounded search, read by a cross-family machine check on 2026-10-26.","canonicalUrl":"https://openai.com/index/stargate-advances-with-partnership-with-oracle/","evidenceUrl":"https://openai.com/index/stargate-advances-with-partnership-with-oracle/","language":"en","archivedUrl":"https://web.archive.org/web/20250722115119/https://openai.com/index/stargate-advances-with-partnership-with-oracle/","archivedAt":"2025-07-22","accessedAt":"2026-09-19","reuse":{"attribution":"OpenAI, Stargate advances with 4.5 GW partnership with Oracle (2025-07-22).","licence":"No open redistribution licence asserted","note":"Bibliographic metadata and editorial paraphrases only; no full source redistributed."}},{"id":"source-eoc-1997-chapter-5","title":"Engines of Creation, chapter 5: Thinking Machines (web edition)","authors":["K. Eric Drexler"],"publisher":"Foresight Institute (web edition of Engines of Creation)","publicationDate":"1986","version":"Web edition page as captured by the Internet Archive on 1997-07-09 (timestamp 19970709112142); it carries the notice \"Copyright 1986, K. Eric Drexler\". Retrieved 2026-10-06.","canonicalUrl":"http://www.foresight.org/EOC/EOC_Chapter_5.html","evidenceUrl":"https://web.archive.org/web/19970709112142/http://www.foresight.org:80/EOC/EOC_Chapter_5.html","language":"en","archivedUrl":"https://web.archive.org/web/19970709112142/http://www.foresight.org:80/EOC/EOC_Chapter_5.html","archivedAt":"1997-07-09","accessedAt":"2026-10-06","reuse":{"attribution":"K. Eric Drexler, Engines of Creation: The Coming Era of Nanotechnology (1986), Foresight Institute web edition.","licence":"All rights reserved, according to the page’s copyright notice.","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"source-extropians-1996-09-04-bostrom","title":"Re: Thinking about the future... (Extropians mailing list)","authors":["Nick Bostrom"],"publisher":"Extropians mailing list","publicationDate":"1996-09-04","version":"Archive HTML as served, retrieved 2026-10-06 with the hash recorded by the 1996–2022 candidate search; quoted lines excluded when reading the author’s text. Archive at extropians.weidai.com. Date: the archive’s receipt line reads “Wed Sep  4 10:09:44 1996 MST” and the message is filed in the archive’s 1996 volume, which give the record’s date; the sent header is garbled and ends “04 Sep 95”, which nothing else in the message supports (receipt line re-read in a direct retrieval on 2026-10-27).","canonicalUrl":"http://extropians.weidai.com/extropians.96/0764.html","evidenceUrl":"http://extropians.weidai.com/extropians.96/0764.html","language":"en","archivedUrl":"https://web.archive.org/web/20150427180540/http://extropians.weidai.com/extropians.96/0764.html","archivedAt":"2015-04-27","accessedAt":"2026-10-06","reuse":{"attribution":"Nick Bostrom, \"Re: Thinking about the future...\", Extropians mailing list, 1996-09-04.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"source-sl4-2000-12-12-yudkowsky-0012-0381","title":"Re: Revising a Friendly AI (SL4 mailing list)","authors":["Eliezer Yudkowsky"],"publisher":"SL4 mailing list","publicationDate":"2000-12-12","version":"Archive HTML as served, retrieved 2026-10-07 (SHA-256 5841a35e50411bec7f05f3bd73189ebc3f4e0408d7e43cc3ed3e6a5551f21176); quoted lines excluded when reading the author’s text. The served archive shows later URL rewriting and does not authenticate historical bytes. Archive at sl4.org.","canonicalUrl":"http://sl4.org/archive/0012/0381.html","evidenceUrl":"http://sl4.org/archive/0012/0381.html","language":"en","archivedUrl":"https://web.archive.org/web/20250907043122/http://sl4.org/archive/0012/0381.html","archivedAt":"2025-09-07","accessedAt":"2026-10-07","reuse":{"attribution":"Eliezer Yudkowsky, “Re: Revising a Friendly AI”, SL4 mailing list, 2000-12-12.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"source-sl4-2001-02-07-yudkowsky-0102-0545","title":"Re: Learning to be evil (SL4 mailing list)","authors":["Eliezer Yudkowsky"],"publisher":"SL4 mailing list","publicationDate":"2001-02-07","version":"Archive HTML as served, retrieved 2026-10-07 (SHA-256 8b0b5d5e8516ba9d0d801d1e8c411d4df7eef1087058b9c022884eaca09d5440); quoted lines excluded when reading the author’s text. The served archive shows later URL rewriting and does not authenticate historical bytes. Archive at sl4.org.","canonicalUrl":"http://sl4.org/archive/0102/0545.html","evidenceUrl":"http://sl4.org/archive/0102/0545.html","language":"en","archivedUrl":"https://web.archive.org/web/20240418072323/http://sl4.org/archive/0102/0545.html","archivedAt":"2024-04-18","accessedAt":"2026-10-07","reuse":{"attribution":"Eliezer Yudkowsky, “Re: Learning to be evil”, SL4 mailing list, 2001-02-07.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"source-eu-digital-omnibus-ai-2026","title":"Regulation (EU) 2026/1744 of the European Parliament and of the Council of 8 July 2026 amending Regulations (EU) 2024/1689, (EU) 2018/1139 and (EU) 2023/1230 as regards the simplification of the implementation of harmonised rules on artificial intelligence (Digital Omnibus on AI)","authors":["European Parliament and Council of the European Union"],"publisher":"Official Journal of the European Union (EUR-Lex)","publicationDate":"2026-07-24","version":"Official Journal L series edition of 24.7.2026, English HTML, retrieved 2026-10-10. Its text was identical to that of the Internet Archive capture of 2026-07-28.","canonicalUrl":"https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=OJ:L_202601744","evidenceUrl":"https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=OJ:L_202601744","language":"en","archivedUrl":"https://web.archive.org/web/20260728193314/https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=OJ:L_202601744","archivedAt":"2026-07-28","accessedAt":"2026-10-10","reuse":{"attribution":"European Parliament and Council of the European Union, Regulation (EU) 2026/1744 (Digital Omnibus on AI), Official Journal L, 24.7.2026.","licence":"Official legal text; no redistribution licence asserted","note":"Selekoa practice: metadata, attributed paraphrases and short attributed quotations under the editorial policy (at most 50 words each; at most 400 quoted words per document across the dataset, and about 10% of a document under about 2,000 words). Documents are linked externally, not redistributed. Publisher terms that forbid quotation override this practice."}},{"id":"source-hanson-economic-growth-given-machine-intelligence","title":"Economic Growth Given Machine Intelligence","authors":["Robin Hanson"],"publisher":"Working paper","publicationDate":null,"version":"Wayback capture of http://hanson.gmu.edu/aigrow.pdf (PDF created 2000-06-22). Capture retrieved 2026-10-08 (SHA-256 754380e21b0e9f5a62fe71ea238a86e8640df48a0551c55cf4877a2a732b99c1). Versions: claimed 1998 version (1998 (author's citation)): not located; hanson.gmu.edu/aigrow.pdf (undated; dated by the 2000-08-19 capture): checked. Published version: Hanson, R. (n.d.). Economic growth given machine intelligence [Unpublished manuscript]. Retrieved from a 2000-08-19 Wayback capture of hanson.gmu.edu/aigrow.pdf. No published version located (Crossref search 2026-10-10 found none).","canonicalUrl":"https://mason.gmu.edu/~rhanson/aigrow.pdf","evidenceUrl":"https://web.archive.org/web/20000819034042/http://hanson.gmu.edu/aigrow.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20000819034042/http://hanson.gmu.edu/aigrow.pdf","archivedAt":"2000-08-19","accessedAt":"2026-10-08","reuse":{"attribution":"Robin Hanson, “Economic Growth Given Machine Intelligence”, Working paper, author's site (hanson.gmu.edu/aigrow.pdf).","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"source-erdil-besiroglu-2023-explosive-growth-review","title":"Explosive growth from AI automation: A review of the arguments","authors":["Ege Erdil","Tamay Besiroglu"],"publisher":"arXiv:2309.11690 [econ.GN]","publicationDate":"2023-10-01","version":"Wayback capture of arxiv.org/pdf/2309.11690v2; compared with v3. Capture retrieved 2026-10-08 (SHA-256 28610b91d03d1d055d6e1af308b964387905af66557936ee49c622f693985617). Versions: arXiv v1 (2023-09-20): not retrieved; arXiv v2 (2023-10-01): checked; arXiv v3 (2024-07-15): compared. Published version: Erdil, E., & Besiroglu, T. (2023). Explosive growth from AI automation: A review of the arguments (arXiv:2309.11690v2) [Preprint]. arXiv. https://arxiv.org/abs/2309.11690v2 No journal version located (Crossref search 2026-10-10 found none).","canonicalUrl":"http://arxiv.org/pdf/2309.11690v2","evidenceUrl":"https://web.archive.org/web/20250630102426/http://arxiv.org/pdf/2309.11690v2","language":"en","archivedUrl":"https://web.archive.org/web/20250630102426/http://arxiv.org/pdf/2309.11690v2","archivedAt":"2025-06-30","accessedAt":"2026-10-08","reuse":{"attribution":"Ege Erdil, Tamay Besiroglu, “Explosive growth from AI automation: A review of the arguments”, arXiv:2309.11690 [econ.GN], 2023-10-01.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"source-mansfield-1966-technological-change","title":"Technological Change: Measurement, Determinants, and Diffusion","authors":["Edwin Mansfield"],"publisher":"In The Employment Impact of Technological Change, Appendix Volume II (U.S. GPO, February 1966)","publicationDate":"1966-02","version":"Internet Archive OCR text (djvu.txt) of a scanned copy, retrieved 2026-10-09 from https://dn710704.ca.archive.org/0/items/ERICED022857/ERICED022857djvu.txt; OCR errors possible; quotes checked only against clean stretches. Current copy; no earlier capture compared. SHA-256 85b1fb59b997e8cbc0f7255076a55f68008955a2b10523b375ac0ae58628572c.","canonicalUrl":"https://archive.org/details/ERIC_ED022857","evidenceUrl":"https://archive.org/details/ERIC_ED022857","language":"en","accessedAt":"2026-10-09","reuse":{"attribution":"Edwin Mansfield, “Technological Change: Measurement, Determinants, and Diffusion”, In The Employment Impact of Technological Change, Appendix Volume II (U.S. GPO, February 1966), 1966-02.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"source-hanson-2008-ai-go-foom-miri-2013","title":"AI Go Foom","authors":["Robin Hanson"],"publisher":"Machine Intelligence Research Institute, The Hanson-Yudkowsky AI-Foom Debate (2013)","publicationDate":"2013","version":"Chapter 10 of The Hanson-Yudkowsky AI-Foom Debate (Machine Intelligence Research Institute, 2013), which compiles the 2008 Overcoming Bias debate with later posts; the compilation heads the post with its date. Current MIRI PDF, retrieved 2026-10-11. SHA-256 6b6dfd0416402e1776d7ddcece95298df2df0ed9bc0bd4a6f4bd81c4b1e5102a.","canonicalUrl":"https://intelligence.org/files/AIFoomDebate.pdf","evidenceUrl":"https://intelligence.org/files/AIFoomDebate.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20260929131419/https://intelligence.org/files/AIFoomDebate.pdf","archivedAt":"2026-09-29","accessedAt":"2026-10-11","reuse":{"attribution":"Robin Hanson, “AI Go Foom”, in Machine Intelligence Research Institute, The Hanson-Yudkowsky AI-Foom Debate (2013).","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"lg-src-deal-adept-update-2024","title":"An update from Adept","authors":["Adept Team"],"publisher":"Adept","publicationDate":"2024-06-28","version":"Dated page, retrieved on 2026-10-15 (SHA-256 857d9acc52fb3134c989d0a1d9a680a016ec30bbc9124ba7f006113282abf726) through the third-party reader relay r.jina.ai, which renders the publisher’s current page (the publisher refused direct retrieval); not an archive copy.","canonicalUrl":"https://www.adept.ai/blog/adept-update","evidenceUrl":"https://www.adept.ai/blog/adept-update","language":"en","archivedUrl":"https://web.archive.org/web/20260813131024/https://www.adept.ai/blog/adept-update/","archivedAt":"2026-08-13","accessedAt":"2026-10-15","reuse":{"attribution":"Adept Team, “An update from Adept”, 2024-06-28.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"lg-src-mx-bi-kroiss","title":"Another Co-Founder Is Exiting Elon Musk's XAI","authors":["Grace Kay"],"publisher":"Business Insider","publicationDate":"2026-03-25","version":"Dated page, retrieved on 2026-10-15 (SHA-256 6ca85271f8ae9a067db52f8ffdfdf07f859ee1e1e2238d06fa29e878997eaa0f) directly from the publisher.","canonicalUrl":"https://www.businessinsider.com/manuel-kroiss-xai-cofounder-departure-elon-musk-2026-3","evidenceUrl":"https://www.businessinsider.com/manuel-kroiss-xai-cofounder-departure-elon-musk-2026-3","language":"en","archivedUrl":"https://web.archive.org/web/20260808024547/https://www.businessinsider.com/manuel-kroiss-xai-cofounder-departure-elon-musk-2026-3","archivedAt":"2026-08-08","accessedAt":"2026-10-15","reuse":{"attribution":"Grace Kay, “Another Co-Founder Is Exiting Elon Musk's XAI”, 2026-03-25.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"lg-src-mx-cnbc-msl-memo","title":"Mark Zuckerberg announces creation of Meta Superintelligence Labs. Read the memo","authors":["CNBC"],"publisher":"CNBC","publicationDate":"2025-06-30","version":"Dated page, retrieved on 2026-10-15 (SHA-256 3c940666760a1f8148ddb1674b0cd4c5f8d28a5eaa32f79fc7052c8cce1ca6b4) through the third-party reader relay r.jina.ai, which renders the publisher’s current page (the publisher refused direct retrieval); not an archive copy.","canonicalUrl":"https://www.cnbc.com/2025/06/30/mark-zuckerberg-creating-meta-superintelligence-labs-read-the-memo.html","evidenceUrl":"https://www.cnbc.com/2025/06/30/mark-zuckerberg-creating-meta-superintelligence-labs-read-the-memo.html","language":"en","archivedUrl":"https://web.archive.org/web/20261015225529/https://www.cnbc.com/2025/06/30/mark-zuckerberg-creating-meta-superintelligence-labs-read-the-memo.html","archivedAt":"2026-10-15","accessedAt":"2026-10-15","reuse":{"attribution":"CNBC, “Mark Zuckerberg announces creation of Meta Superintelligence Labs. Read the memo”, 2025-06-30.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"lg-src-new-mistral-about","title":"About Mistral AI (company page, undated)","authors":["Mistral AI"],"publisher":"Mistral AI","publicationDate":null,"version":"Undated page, retrieved on 2026-10-15 (SHA-256 89e4c4c3d1c20eeb8820643698f5f39656afac71b6eb367ed1cb5141094d6dec) directly from the publisher. Its wording can change; earlier wording is not authenticated.","canonicalUrl":"https://mistral.ai/about","evidenceUrl":"https://mistral.ai/about","language":"en","archivedUrl":"https://web.archive.org/web/20261015151143/https://mistral.ai/about/","archivedAt":"2026-10-15","accessedAt":"2026-10-15","reuse":{"attribution":"Mistral AI, “About Mistral AI (company page, undated)”.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"source-g7-hiroshima-code-of-conduct-2023","title":"Hiroshima Process International Code of Conduct for Organizations Developing Advanced AI Systems","authors":["G7"],"publisher":"Ministry of Internal Affairs and Communications, Japan (Hiroshima AI Process site)","publicationDate":null,"version":"Retrieved on 2026-10-27 (SHA-256 14cfe36633502daa48231b49ca7f186c53a72add181b328047665763ecb13ac4). Undated English PDF; dated by the G7 Leaders' Statement of 30 October 2023, which attaches it","canonicalUrl":"https://www.soumu.go.jp/hiroshimaaiprocess/pdf/document05_en.pdf","evidenceUrl":"https://www.soumu.go.jp/hiroshimaaiprocess/pdf/document05_en.pdf","language":"en","accessedAt":"2026-10-27","reuse":{"attribution":"G7, “Hiroshima Process International Code of Conduct for Organizations Developing Advanced AI Systems”.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"source-cpo-ch-mofcom-ree-61-2025","title":"商务部公告2025年第61号 公布对境外相关稀土物项实施出口管制的决定","authors":["商务部"],"publisher":"中华人民共和国商务部（安全与管制局）/ Ministry of Commerce of the PRC, page on aqygzj.mofcom.gov.cn","publicationDate":"2025-10-09","version":"Retrieved for the candidate pack on 2026-10-28 (SHA-256 f22e765ad9d4b619c5170f2fe1235aa90e9ab8d833e66812c7b69a64effdedb6). Announcement as served on the MOFCOM Bureau of Industry Security subsite on 2026-10-28; dated 2025年10月9日. Retrieved again at admission on 2026-10-30 (SHA-256 9fa717248dc38944358cf220f6abb02255d7cc33515cb696cf7f8c12ffb1cbb9).","canonicalUrl":"https://aqygzj.mofcom.gov.cn/qdml/art/2025/art_fbc5f7e23bff4bb184d5c49ff878f8ad.html","evidenceUrl":"https://aqygzj.mofcom.gov.cn/qdml/art/2025/art_fbc5f7e23bff4bb184d5c49ff878f8ad.html","language":"zh-Hans","accessedAt":"2026-10-28","reuse":{"attribution":"商务部, “商务部公告2025年第61号 公布对境外相关稀土物项实施出口管制的决定”, 2025-10-09.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"source-ab-batalis-cset-2023","title":"AI and Biorisk: An Explainer (extended explainer PDF)","authors":["Center for Security and Emerging Technology, Georgetown University"],"publisher":"Center for Security and Emerging Technology, Georgetown University","publicationDate":null,"version":"Retrieved for the candidate pack on 2026-10-28 (SHA-256 b42589d1df9f0b702975591dc9c064b7d92a9c761e9a24b1681668055af29c05). Extended explainer PDF, file name marked FINAL-v2, as served on 2026-10-28 Retrieved again at admission on 2026-10-30 (SHA-256 b42589d1df9f0b702975591dc9c064b7d92a9c761e9a24b1681668055af29c05, the same bytes).","canonicalUrl":"https://cset.georgetown.edu/wp-content/uploads/20230047-AI-BioRisk-Extended-Explainer-FINAL-v2.pdf","evidenceUrl":"https://cset.georgetown.edu/wp-content/uploads/20230047-AI-BioRisk-Extended-Explainer-FINAL-v2.pdf","language":"en","accessedAt":"2026-10-28","reuse":{"attribution":"Center for Security and Emerging Technology, Georgetown University, “AI and Biorisk: An Explainer (extended explainer PDF)”.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"source-ab-cset-biorisk-explainer","title":"AI and Biorisk: An Explainer (publication page)","authors":["Center for Security and Emerging Technology, Georgetown University"],"publisher":"Center for Security and Emerging Technology, Georgetown University","publicationDate":null,"version":"Retrieved for the candidate pack on 2026-10-28 (SHA-256 b8deadfa4acfb2e37de06cc19e3ae110b086cd557207350c772b58e8ca75fa13). Publication page as served on 2026-10-28; used only for the publication month and the author's name Retrieved again at admission on 2026-10-30 (SHA-256 61f2fd18470acc7fbfe139699cfce023887e357f419943a5c8863fac0192412a). Its wording can change; earlier wording is not authenticated.","canonicalUrl":"https://cset.georgetown.edu/publication/ai-and-biorisk-an-explainer/","evidenceUrl":"https://cset.georgetown.edu/publication/ai-and-biorisk-an-explainer/","language":"en","accessedAt":"2026-10-28","reuse":{"attribution":"Center for Security and Emerging Technology, Georgetown University, “AI and Biorisk: An Explainer (publication page)”.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}}],"evidence":[{"id":"evidence-rsp-v3-2-reviewer-approval","sourceRole":"primary","support":"supports","locator":"PDF p. 20 (printed p. 19), Changelog, April 29, 2026 (RSP v3.2) entry","passage":"This update authorizes the LTBT to request external review of Risk Reports, gives the LTBT the power to approve our selection of external reviewers, and formalizes a requirement that we provide the LTBT with regular briefings.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-03","note":"Verbatim excerpt checked by an automated research assistant on 2026-10-03 against the inspected PDF, byte-identical to the Internet Archive capture of 2026-05-05; it replaces an earlier attributed paraphrase. Human editorial review is pending."},{"id":"evidence-openai-oracle-capacity-2025","sourceRole":"primary","support":"supports","locator":"Opening paragraph","passage":"Oracle and OpenAI have entered an agreement to develop 4.5 gigawatts of additional Stargate data center capacity in the U.S.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-26","note":"Re-checked 2026-10-15 (research-debt pass resolving Audit B's RESEARCH MORE). openai.com refused direct retrieval (HTTP 403) and the Wayback Machine did not answer; the quotation was verified verbatim against the current page text obtained through a reader relay (r.jina.ai), not against a dated capture, so it does not authenticate the July 2025 bytes. Company statements are not independent proof of implementation. Human source review pending. Quotation extended through “in the U.S.” by the maintainer’s decision of 2026-10-26, after a cross-family machine check read the full sentence in the Wayback Machine capture of 2025-07-22 11:51:19 UTC on 2026-10-26 (research/archive-checks-2026-10-26/codex-report.md, item 6); that capture could not be re-fetched when the change was made. It records an agreement to develop capacity, not completed or operational capacity."},{"id":"evidence-prov-rsp-v2-1-misuse-uplift-threshold","sourceRole":"primary","support":"supports","locator":"PDF p. 8 (printed p. 4), §2 Capability Thresholds and Required Safeguards, table row 'Chemical, Biological, Radiological, and Nuclear (CBRN) weapons', entry 'CBRN-3', Capability Thresholds column","passage":"The ability to significantly help individuals or groups with basic technical backgrounds (e.g., undergraduate STEM degrees) create/obtain and deploy CBRN weapons.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-03","note":"Verbatim excerpt checked by an automated research assistant on 2026-10-03 against the inspected PDF, byte-identical to the Internet Archive capture of 2025-04-02. Extraction note: Wording as in v2.0; v2.1 adds the label 'CBRN-3' and a new CBRN-4 threshold in the same row (not proposed: quotation limit). Human editorial review is pending."},{"id":"evidence-ts-a5-drexler-1986","sourceRole":"primary","support":"supports","locator":"Chapter 5, \"Accelerating the Technology Race\", final synthesis","passage":"At some point, full-fledged automated engineering systems will pull ahead on their own. [...] Then assembler-built AI systems will bring still swifter automated engineering, evolving technological ideas at a pace set by systems a million times faster than a human brain.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-06","note":"The same passage as the A1 record."},{"id":"evidence-ts-a3-drexler-1986-context","sourceRole":"primary","support":"supports","locator":"Chapter 5, \"Accelerating the Technology Race\", final synthesis","passage":"Molecular technology, developed with the help of automated engineering, lets assembler-built AI systems bring \"still swifter automated engineering\".","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-06","note":"Checked against the 1997 capture."},{"id":"evidence-ts-a3-bostrom-1996-09-04","sourceRole":"primary","support":"supports","locator":"Message body, first unquoted paragraph of the reply","passage":"think out a radical improvement to its design; which would make it more intelligent, allowing it to accelerate further","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-06","note":"Checked against the archive HTML on 2026-10-06."},{"id":"evidence-yk-t30-s2","sourceRole":"primary","support":"supports","locator":"message body, paragraph beginning 'Anyhoo, Friendliness isn't intended to suppress…'","passage":"AIs don't have 'em unless you put them there. Correspondingly, although other possibilities exist, the default, engineering-*conservative* assumption is that goodness also doesn't materialize in source code.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-07","note":"Candidate pack passage T30-s2. The quotation (27 words) matched the retrieved expression (SHA-256 8b0b5d5e8516ba9d0d801d1e8c411d4df7eef1087058b9c022884eaca09d5440) when checked on 2026-10-07."},{"id":"evidence-omni-enacted-1","sourceRole":"primary","support":"supports","locator":"Enacting formula before the recitals","passage":"HAVE ADOPTED THIS REGULATION:","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-10","note":"Verbatim quote, matched mechanically (after whitespace normalisation) by the proposing session against the text retrieved on 2026-10-10. Checked by fresh-context machine audit on 2026-10-10 (review/frontier-governance-audit-2026-10-10.md). Human editorial review is pending."},{"id":"evidence-omni-enacted-2","sourceRole":"primary","support":"supports","locator":"Official Journal header","passage":"L series 2026/1744 24.7.2026","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-10","note":"Verbatim quote, matched mechanically (after whitespace normalisation) by the proposing session against the text retrieved on 2026-10-10. Gives the publication date 24.7.2026. Checked by fresh-context machine audit on 2026-10-10 (review/frontier-governance-audit-2026-10-10.md). Human editorial review is pending."},{"id":"evidence-omni-enacted-3","sourceRole":"primary","support":"supports","locator":"Closing formula","passage":"Done at Strasbourg, 8 July 2026.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-10","note":"Verbatim quote, matched mechanically (after whitespace normalisation) by the proposing session against the text retrieved on 2026-10-10. Checked by fresh-context machine audit on 2026-10-10 (review/frontier-governance-audit-2026-10-10.md). Human editorial review is pending."},{"id":"evidence-omni-enacted-4","sourceRole":"primary","support":"context","locator":"Entry into force: Article 4, first paragraph","passage":"This Regulation shall enter into force on the third day following that of its publication in the Official Journal of the European Union.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-10","note":"Verbatim quote, matched mechanically (after whitespace normalisation) by the proposing session against the text retrieved on 2026-10-10. Context for the effective date recorded on the edition (2026-07-27), not for the adoption itself. Checked by fresh-context machine audit on 2026-10-10 (review/frontier-governance-audit-2026-10-10.md). Human editorial review is pending."},{"id":"evidence-ec-g04-e","sourceRole":"primary","support":"supports","locator":"Section 4 \"Some Variations\", paragraph on exclusively human jobs","passage":"Having some exclusively human jobs slows growth somewhat.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-14","note":"Candidate pack passage G04-e (research/economics-growth). The quotation (8 words) matched the checked expression (hanson-aigrow) in the pack's verification. Re-checked at admission on 2026-10-14 against a copy retrieved that day.","contextOnly":{"reason":"Alone, \"slows growth somewhat\" hides that growth in the example remains explosive.","requires":"The next sentence of the paper with the 27-month doubling time."}},{"id":"evidence-ec-g05-g","sourceRole":"primary","support":"supports","locator":"Section 3.2 \"Output is bottlenecked by other non-accumulable factors of production\", paragraph containing \"there might be intrinsic timescales\"","passage":"Some version of this argument is certainly sound: there must eventually be some resource constraints that prevent output from growing arbitrarily large. The important question about this argument is not whether it holds eventually, but whether it holds quickly enough to preclude explosive growth.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-14","note":"Candidate pack passage G05-g (research/economics-growth). The quotation (44 words) matched the checked expression (erdil-arxiv-v2) in the pack's verification. Re-checked at admission on 2026-10-14 against a copy retrieved that day."},{"id":"evidence-ec-e01-08","sourceRole":"primary","support":"supports","locator":"\"AI Go Foom\", 20 November 2008, MIRI compilation (2013) ch. 10, post body; pdf p. 63","passage":"The growth rate of its “intelligence” may be slow when it is dumb, but gets faster as it gets smarter.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-29","note":"Candidate pack passage E01-08 (research/economics-mainstream/esoteric). The quotation (20 words) matched the retrieved current copy (SHA-256 6b6dfd0416402e1776d7ddcece95298df2df0ed9bc0bd4a6f4bd81c4b1e5102a) in the pack's verification of 2026-10-15. Re-checked at admission on 2026-10-16 against a copy retrieved that day. On 2026-10-30, by the maintainer’s decision of that day, the quotation was corrected to the archive capture of 2026-09-29 (quotation marks and apostrophes as in the capture), which a cross-family archive check read on 2026-10-29; the proposition is unchanged.","contextOnly":{"reason":"Fresh-context audit (A1) judged that the quotation misleads when shown alone.","requires":"Hanson is restating Yudkowsky's argument ('I guess it is up to me to summarize Eliezer's argument as best I can, so I can then respond'), ending 'QED. So, Eliezer, is this close enough...?'. This is not Hanson's own view; he rejects it in the debate."}},{"id":"lg-ev-new-mistral-founding-2023","sourceRole":"primary","support":"supports","locator":"Section “Our frontier lab roots”","passage":"The page states that Mistral was born in April 2023.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The paraphrase’s verbatim anchor “April 2023, Mistral was born” matched the retrieved text on 2026-10-15 and again in a fresh retrieval the same day. Read directly from the publisher. Matching ignored differences of whitespace, case, quotation marks, apostrophes, dashes and Unicode compatibility forms, so the source’s typography can differ. The paraphrase is Selekoa’s; only the anchor is the source’s wording. No human editorial review.","retrievalRoute":"direct"},{"id":"lg-ev-deal-brock-adept-ceo-2024","sourceRole":"primary","support":"supports","locator":"Paragraph 6","passage":"Zach Brock, Head of Engineering, will be taking over as Adept’s CEO.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The quotation (12 words) matched the retrieved text on 2026-10-15 and again in a fresh retrieval the same day. Read through the third-party reader relay r.jina.ai, which renders the publisher’s current page (the publisher refused direct retrieval); not an archive copy. Matching ignored differences of whitespace, case, quotation marks, apostrophes, dashes and Unicode compatibility forms, so the source’s typography can differ. No human editorial review.","retrievalRoute":"reader_relay"},{"id":"lg-ev-mx-bansal-msl-2025","sourceRole":"secondary","support":"supports","locator":"Memo, list of new team members","passage":"The memo describes Trapit Bansal as having pioneered RL on chain of thought and as a co-creator of OpenAI's o-series models.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The paraphrase’s verbatim anchor “Trapit Bansal” matched the retrieved text on 2026-10-15 and again in a fresh retrieval the same day. Read through the third-party reader relay r.jina.ai, which renders the publisher’s current page (the publisher refused direct retrieval); not an archive copy. Matching ignored differences of whitespace, case, quotation marks, apostrophes, dashes and Unicode compatibility forms, so the source’s typography can differ. The paraphrase is Selekoa’s; only the anchor is the source’s wording. A cross-family machine check (2026-10-16/17) re-read the record’s sources and agreed with the record. No human editorial review. Reworded on 2026-10-29 to say “describes … as” (consistency audit fix A4); what the paraphrase reports is unchanged.","retrievalRoute":"reader_relay"},{"id":"lg-ev-mx-yang-cofounded-xai","sourceRole":"secondary","support":"supports","locator":"Paragraphs 2–3 (the sentence ending “now sits at 10” closes paragraph 2; the list of names opens paragraph 3)","passage":"the number of cofounder departures now sits at 10. Guodong Zhang, Zihang Dai, Toby Pohlen, Jimmy Ba, Tony Wu, and Greg Yang have all stepped away since January.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The quotation (28 words) matched the retrieved text on 2026-10-15 and again in a fresh retrieval the same day. Read directly from the publisher. Matching ignored differences of whitespace, case, quotation marks, apostrophes, dashes and Unicode compatibility forms, so the source’s typography can differ. A cross-family machine check (2026-10-16/17) re-read the record’s sources and agreed with the record. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-ap-edwin-mansfield-mansfield-1966-technological-change","sourceRole":"primary","support":"supports","locator":"Contribution title page, “Prepared for the Commission by” (OCR text of the scanned volume)","passage":"Edwin Mansfield","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-27","note":"Passage and locator as inspected by a reviewer from a different model family on 2026-10-27 in a direct retrieval of the Internet Archive scan’s OCR text; the preparer’s matcher had found the name only outside the byline; the page image was not inspected. Human editorial review is pending.","retrievalRoute":"direct"},{"id":"evidence-ap-eliezer-yudkowsky-sl4-2000-12-12-yudkowsky-0012-0381","sourceRole":"primary","support":"supports","locator":"Message header, sender name (address not reproduced)","passage":"Eliezer S. Yudkowsky","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-27","note":"Credited name checked by an automated research assistant on 2026-10-27 in a direct retrieval. Human editorial review is pending.","retrievalRoute":"direct"},{"id":"ev-cmp-fsf-v3-1-d5-cadence-1","sourceRole":"primary","support":"supports","locator":"PDF p. 5 (printed p. 5), §1.3.2 Inherent Risk Assessment, first paragraph, second sentence, and footnote 4","passage":"Google DeepMind conducts a critical capability assessment before a new frontier AI model’s first external deployment; under footnote 4, one may be skipped for a low-risk external deployment if the appropriate governance function finds its residual risk acceptable, even at a reached T/CCL (with a safety case at a CCL).","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-28","note":"Explicit paraphrase by an automated research assistant, corrected on 2026-10-28 after a blind check of all 287 comparison paraphrases by a reviewer from a different model family on 2026-10-28 (research/frontier-framework-comparison/codex-paraphrases-2026-10-28.md), which found that it left out footnote 4, attached to the same sentence, under which the assessment may not be conducted for low-risk external deployments. The correction was written from the passage as re-read in the publisher copy retrieved on 2026-10-28, whose SHA-256 matches the copy retrieved on 2026-10-15. Given as a paraphrase to keep the edition within the quotation limit. The footnote’s example of a low-risk external deployment is release to a few trusted testers. Human editorial review is pending."},{"id":"ev-cmp-faif-2026-d7-internal-1","sourceRole":"primary","support":"supports","locator":"PDF p. 2, §2 Risk Management Framework, third paragraph","passage":"The edition says the results of xAI’s processes and measures for assessing and mitigating systemic risk will be documented.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-15","note":"Explicit paraphrase, written on 2026-10-27 by an automated research assistant from the comparison’s quotation of this passage, which was checked on 2026-10-15 against the publisher copy (SHA-256 recorded in the comparison pack). The edition could not be retrieved again on 2026-10-27; the archive capture of 2026-07-16 has the same SHA-256 (research/archive-checks-2026-10-26/codex-report.md). Given as a paraphrase to keep the edition within the quotation limit. Human editorial review is pending."},{"id":"evidence-fl-intl-g7-code-testing-1","sourceRole":"primary","support":"supports","locator":"PDF p. 2, Action 1","passage":"This includes employing diverse internal and independent external testing measures, through a combination of methods for evaluations, such as red-teaming, and implementing appropriate mitigation to address identified risks and vulnerabilities.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-28","note":"Candidate pack passage, frameworks to law (research/frameworks-to-law). The quotation (30 words) matched the text retrieved for the candidate pack on 2026-10-27 and again at admission on 2026-10-28, in a fresh retrieval whose SHA-256 equals the pack’s. Read directly from www.soumu.go.jp. Matching ignored differences of whitespace, case, quotation marks, dashes and Unicode compatibility forms, so the source’s typography can differ. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-cpo-ch-cn-mofcom-2025-61-issued-1","sourceRole":"primary","support":"supports","locator":"Article 1","passage":"必须获得中国商务部颁发的两用物项出口许可证件","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-30","note":"Candidate pack passage, compute, power and orbit (research/compute-power-orbit). The quotation (22 words, each CJK character counted as one) matched the text retrieved for the candidate pack on 2026-10-28 and again at admission on 2026-10-30, in a fresh retrieval (SHA-256 9fa717248dc38944358cf220f6abb02255d7cc33515cb696cf7f8c12ffb1cbb9; the bytes differ from the pack’s retrieval). Matching ignored differences of whitespace, case, quotation marks, dashes and Unicode compatibility forms, so the source’s typography can differ. English machine translation, not verified by a person: “must obtain a dual-use item export licence issued by the Chinese Ministry of Commerce”. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-cpo-ch-cn-mofcom-2025-61-issued-2","sourceRole":"primary","support":"supports","locator":"Article 8","passage":"parts 1(1) and 1(2) take effect on 1 December 2025 and part 1(3) on publication.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-30","note":"Candidate pack passage, compute, power and orbit (research/compute-power-orbit). The paraphrase is Selekoa’s, written from the text retrieved for the candidate pack on 2026-10-28; it is not a quotation and was not machine-matched. The cross-family check found it faithful. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-ab-arg-d2-batalis-context-1","sourceRole":"primary","support":"supports","locator":"Section on scientifically naive users, PDF p. 6","passage":"In sum, the barrier to scientific information is not high, even for beginners. However, the feasibility of actually producing these agents depends on having both information and hands-on production success.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-30","note":"Candidate pack passage, AI and biosecurity (research/ai-bio). The quotation (30 words) matched the text retrieved for the candidate pack on 2026-10-28 and again at admission on 2026-10-30, in a fresh retrieval whose SHA-256 equals the pack’s. Matching ignored differences of whitespace, case, quotation marks, dashes and Unicode compatibility forms, so the source’s typography can differ. The explainer is the primary source for its own statement. A cross-family machine check (2026-10-30, item 107) re-read the record’s sources and found full support. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-ab-arg-d2-batalis-context-2","sourceRole":"primary","support":"supports","locator":"Key takeaway 1, PDF p. 1","passage":"Biorisk is already possible without AI, even for non-experts. AI tools are not needed to access the foundational information and resources to cause biological harm.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-30","note":"Candidate pack passage, AI and biosecurity (research/ai-bio). The quotation (25 words) matched the text retrieved for the candidate pack on 2026-10-28 and again at admission on 2026-10-30, in a fresh retrieval whose SHA-256 equals the pack’s. Matching ignored differences of whitespace, case, quotation marks, dashes and Unicode compatibility forms, so the source’s typography can differ. The same point at the level of the explainer's own key takeaway. A cross-family machine check (2026-10-30, item 107) re-read the record’s sources and found full support. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-ab-arg-d2-batalis-context-3","sourceRole":"primary","support":"supports","locator":"Publication page, title block: the author and date line beneath the heading \"AI and Biorisk: An Explainer\"","passage":"Steph Batalis December 2023","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-30","note":"Candidate pack passage, AI and biosecurity (research/ai-bio). The quotation (4 words) matched the text retrieved for the candidate pack on 2026-10-28 and again at admission on 2026-10-30, in a fresh retrieval (SHA-256 61f2fd18470acc7fbfe139699cfce023887e357f419943a5c8863fac0192412a; the bytes differ from the pack’s retrieval). Matching ignored differences of whitespace, case, quotation marks, dashes and Unicode compatibility forms, so the source’s typography can differ. Added after the audit of 2026-10-28: the December 2023 date rested on no cited passage. Cited for the publication month only; the explainer PDF carries no publication date. The page renders the author and the date as adjacent elements beneath the title, which the text extraction joins into one line. A cross-family machine check (2026-10-30, item 107) re-read the record’s sources and found full support. No human editorial review.","retrievalRoute":"direct"}],"at":{"entities":[0,3,19,34,43,46,47,52,57,59,243,269,291,338,340,378,454,458,490,492,507,508,524,533,573,585,591,614,741,812,837,896,1026,1083,1106,1234,1255],"claims":[92,113,406,808,809,870,1188,1210,1268,1278,1484,1643,1705,1770,2014,2286,2392,2631,2689,2931,3059,3423],"sources":[37,41,47,52,58,269,292,421,425,458,475,484,526,539,588,696,700,725,859,917,1070,1071],"evidence":[109,132,438,1018,1019,1084,1441,1482,1483,1484,1485,1586,1599,1851,2071,2159,2277,2603,2933,3039,3303,3377,3789,3998,3999,4719,4720,4721]},"version":"1lg47t7"}