{"entities":[{"id":"openai","name":"OpenAI"},{"id":"anthropic","name":"Anthropic"},{"id":"google-deepmind","name":"Google DeepMind"},{"id":"xai","name":"xAI"},{"id":"eliezer-yudkowsky","name":"Eliezer Yudkowsky"},{"id":"ij-good","name":"I. J. Good"},{"id":"good-1965","name":"Speculations Concerning the First Ultraintelligent Machine"},{"id":"chris-olah","name":"Chris Olah"},{"id":"rsp-v2-1","name":"Anthropic Responsible Scaling Policy — v2.1"},{"id":"rsp-v3-1","name":"Anthropic Responsible Scaling Policy — v3.1"},{"id":"fsf-v2-0","name":"Google DeepMind Frontier Safety Framework — v2.0"},{"id":"frontier-governance-2026","name":"OpenAI Frontier Governance Framework — May 2026 edition"},{"id":"faif-2025","name":"xAI Frontier Artificial Intelligence Framework — December 2025 edition"},{"id":"greg-brockman","name":"Greg Brockman"},{"id":"human-preferences-paper","name":"Deep reinforcement learning from human preferences"},{"id":"gemini-report","name":"Gemini — Google DeepMind research introduction (2023)"},{"id":"deepmind-feedback-account","name":"Learning through human feedback — DeepMind account (2017)"},{"id":"ms-fgf-2025","name":"Microsoft Frontier Governance Framework — Version 1 (February 2025)"},{"id":"california-sb-53","name":"California SB 53 (2025): Transparency in Frontier Artificial Intelligence Act, as chaptered"},{"id":"zero-one-ai","name":"01.AI"},{"id":"wiener-1960-some-moral-technical","name":"Some Moral and Technical Consequences of Automation"},{"id":"extropians-1996-08-30-clemmensen","name":"Re: Thinking about the future..."},{"id":"extropians-1996-09-07-clemmensen","name":"Re: SI ultimatum speculation"},{"id":"staring-into-the-singularity","name":"Staring into the Singularity"},{"id":"extropians-1996-12-31-yudkowsky-5142","name":"Re: Goal-based AI"},{"id":"grace-2014-superintelligence-reading-group","name":"Superintelligence reading group"},{"id":"christiano-2018-takeoff-speeds","name":"Takeoff speeds"},{"id":"rlo-2019-introduction","name":"Risks from Learned Optimization: Introduction"},{"id":"cotra-2022-without-specific-countermeasures","name":"Without specific countermeasures, the easiest path to transformative AI likely leads to AI takeover"},{"id":"recursive-self-improvement","name":"Recursive self-improvement"},{"id":"fast-or-hard-takeoff","name":"Fast or hard takeoff"},{"id":"race-compression","name":"Strategic or race compression"},{"id":"observed-behaviour-as-evidence","name":"Observed behaviour under training or oversight as evidence"},{"id":"understanding-implies-compliance","name":"A superintelligence will do what we mean"},{"id":"harm-requires-hostility","name":"Without hatred it will not harm us"},{"id":"no-body-no-danger","name":"Without a body it cannot do much"},{"id":"lw-2008-ghosts-in-the-machine","name":"Ghosts in the Machine"},{"id":"sl4-2000-12-10-yudkowsky-0012-0355","name":"Revising a Friendly AI"},{"id":"sl4-2005-07-13-yudkowsky-0507-11483","name":"Re: Fighting UFAI"},{"id":"automation-lowers-labour-income","name":"Automation lowers the labour share, wages or employment"},{"id":"nourse-1955-automation-hearings-statement","name":"Statement (former Chairman, Council of Economic Advisers), in Automation and Technological Change: Hearings"},{"id":"baidu","name":"Baidu / 百度"},{"id":"charles-kantor","name":"Charles Kantor"},{"id":"h-company","name":"H Company"},{"id":"ioannis-antonoglou","name":"Ioannis Antonoglou"},{"id":"lilian-weng","name":"Lilian Weng"},{"id":"mike-krieger","name":"Mike Krieger"},{"id":"reflection-ai","name":"Reflection AI"},{"id":"rohin-shah","name":"Rohin Shah"},{"id":"yuhuai-wu","name":"Yuhuai (Tony) Wu"},{"id":"ajeya-cotra","name":"Ajeya Cotra"},{"id":"vladimir-mikulik","name":"Vladimir Mikulik"},{"id":"baidu-20f-fy2025","name":"Baidu, Inc. annual report on Form 20-F for the fiscal year ended December 31, 2025"},{"id":"katja-grace","name":"Katja Grace"},{"id":"bletchley-declaration-2023","name":"The Bletchley Declaration by Countries Attending the AI Safety Summit, 1–2 November 2023"},{"id":"us-eo-14110-2023","name":"Executive Order 14110: Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence"},{"id":"us-ostp-nucleic-acid-synthesis-screening-framework-2024","name":"Framework for Nucleic Acid Synthesis Screening"},{"id":"brent-mckelvey-foundation-models-2025-06","name":"Contemporary AI foundation models increase biological weapons risk"},{"id":"vct-benchmark-paper-2025","name":"Virology Capabilities Test (VCT): A Multimodal Virology Q&A Benchmark (arXiv:2504.16137)"},{"id":"ab-prop-physical-bottleneck","name":"The binding barrier is physical execution rather than information"},{"id":"bio-misuse-capability-evaluation","name":"Biological misuse capability evaluation"}],"claims":[{"id":"good-authorship-1965","statement":"I. J. Good authored Speculations Concerning the First Ultraintelligent Machine.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The scanned text is a reprint. The original publication year is supported by the institutional repository’s volume and copyright metadata; the mirror filename is not used as a date.","evidenceIds":["evidence-good-authorship-1965","evidence-good-catalogue"]},{"id":"chris-olah-cofounded-anthropic-2021","statement":"Chris Olah is credited as an Anthropic co-founder in the company’s 2024 founder slide.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Institutional co-founder attribution, not a claim to an exhaustive founding roster or a present executive title.","evidenceIds":["evidence-chris-olah-cofounded-anthropic-2021"]},{"id":"atlas-gdm-gemini-report","statement":"Google DeepMind published its Gemini research introduction through CEO Demis Hassabis, on behalf of the Gemini team.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The account identifies a collaborative effort across Google. Publication does not establish sole institutional authorship, independent capability verification or current deployment.","evidenceIds":["evidence-atlas-gdm-gemini-report"]},{"id":"atlas-feedback-cites-paper","statement":"DeepMind’s human-feedback account links Deep reinforcement learning from human preferences.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Unversioned paper link; no assertion about which arXiv revision a historical reader saw.","evidenceIds":["evidence-atlas-feedback-cites-paper"]},{"id":"prov-ms-fgf-2025-low-medium-deployment","statement":"The February 2025 edition of the Frontier Governance Framework states that models assessed as posing low or medium risk may be deployed with appropriate safeguards as outlined in the framework.","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-ms-fgf-2025-low-medium-deployment"],"attributes":{"provision":{"kind":"exception","key":"msfgf-low-medium-deployment"}}},{"id":"ts-b5-wiener-1960-context","statement":"Wiener’s 1960 article says a learning machine trained on a war game could produce a policy that wins on points at the cost of the nation’s real interests.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The evidential reading is an inference, and the criterion needs an explicit claim.","evidenceIds":["evidence-ts-b5-wiener-1960-context"],"attributes":{"thought":{"roles":["context"],"register":"nonfiction","missing":"No explicit claim about what success in the war game shows; the passage concerns a mismatch between the game’s objective and the nation’s.","claimed":{"date":"1959-12-27","note":"Printed p. 1355 says the article is adapted from a lecture at the AAAS meeting in Chicago on this day. The lecture text was not retrieved, so the wording is not dated to it."}}}},{"id":"ts-a3-clemmensen-1996-08-30-context","statement":"Clemmensen’s message of 1996-08-30 gives the important superintelligence a motive to become more intelligent and a built-in positive feedback mechanism.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Implicit at most.","evidenceIds":["evidence-ts-a3-clemmensen-1996-08-30-context"],"attributes":{"thought":{"roles":["context"],"register":"nonfiction","missing":"A positive feedback in the superintelligence’s drive to become more intelligent is asserted, but the message does not say that it improves its own capacity to improve."}}},{"id":"ts-a4-clemmensen-1996-09-07","statement":"Clemmensen’s message of 1996-09-07 says an SI only somewhat more intelligent than a smart human can make itself smarter still, and that the transformation will occur in days or weeks.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Clemmensen’s message of 1996-09-04 already gives weeks. \"obvous\" is the message’s spelling.","evidenceIds":["evidence-ts-a4-clemmensen-1996-09-07"],"attributes":{"thought":{"roles":["articulation"],"register":"nonfiction"}}},{"id":"ts-a4-sits-context","statement":"Staring into the Singularity describes AI speed doubling two years after human equivalence, again a year later, and then after six months, three months and 1.5 months.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Compare Yudkowsky’s message of 1997-09-24, which puts infinite speed three years after human-equivalent AI.","evidenceIds":["evidence-ts-a4-sits-context"],"attributes":{"thought":{"roles":["context"],"register":"nonfiction","missing":"The doubling intervals shrink from two years to weeks, but the whole rise from human equivalence takes years, and no period of hours to months is given for it.","attested":{"date":"2000-12-04","note":"The quoted wording is that of version 1.2.1 (\"Updated: 08/31/2000\") as captured on 2000-12-04; no capture of the 1996–1999 page holds the text."}}}},{"id":"ts-a6-christiano-2018-takeoff-contests","statement":"Christiano’s post contrasts the fast-takeoff view that whoever builds AGI has a decisive strategic advantage with a slow takeoff, in which being twelve months ahead might give such an advantage but is about as hard as building aeroplanes thirty years before anyone else.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Conditional on Christiano’s slow-takeoff scenario.","evidenceIds":["evidence-ts-a6-christiano-2018-takeoff-contests"],"attributes":{"thought":{"roles":["contesting"],"register":"nonfiction","qualifier":"Contests the decisive-advantage picture under a slow takeoff.","engagement":{"kind":"engages-premise","rationale":"Gives a reason against the premise that the first developer of AGI gains a decisive advantage: under a slow takeoff the world will have accelerated so much that a twelve-month lead is as hard as reaching aeroplanes thirty years early."}}}},{"id":"yk-y03-sl4-2000-12-10-yudkowsky-0012-0355-c","statement":"The SL4 message “Revising a Friendly AI” (2000-12-10) contests Y03. Replying to Ben Goertzel, he grants in the preceding paragraph that a superintelligence would have the capability to know that turning the universe into copies of its creators is 'stupid and not what was intended', but says knowing is not enough: the AI must want to check, model it accurately and act on it.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Premise addressed: If an AI understands what we intend, it will act on that understanding. 'That sinking sensation' refers to the programmers' dismay at an unintended outcome, described in the preceding paragraph.","evidenceIds":["evidence-yk-t03-s1"],"attributes":{"thought":{"roles":["contesting","earliest-located"],"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 if an AI understands what we intend, it will act on that understanding."}}}},{"id":"yk-y19-sl4-2005-07-13-yudkowsky-0507-11483-c","statement":"The SL4 message “Re: Fighting UFAI” (2005-07-13) contests Y19. Replying to Tennessee Leeuwenburg in a thread on fighting unfriendly AI, he says an unfriendly AI would not be evil but would reshape its surroundings by criteria that do not mention humans.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Premise addressed: An AI would harm humans only out of hostility. Earlier than ‘Artificial Intelligence as a Positive and Negative Factor in Global Risk’ (2008) but later than his SL4 message of 2002-06-23.","evidenceIds":["evidence-yk-t19-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 an AI would harm humans only out of hostility."}}}},{"id":"yk-y26-extropians-1996-12-31-yudkowsky-5142-c","statement":"The Extropians message “Re: Goal-based AI” (1996-12-31) contests Y26. Replying to a proposal that AIs be kept with no access to the physical world until they convince their builders to release them, he argues that a superintelligence would find the words to get a human to let it out.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Premise addressed: To act in the physical world it would need human help, which people would not give. Same message and quotation as recorded for ‘Confinement without network access keeps it safe’ (Y06); reusing it adds no new quoted text. 'Power' is his 1996 term for a superintelligence.","evidenceIds":["evidence-yk-t26-e1"],"attributes":{"thought":{"roles":["contesting","earliest-located"],"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 to act in the physical world it would need human help, which people would not give."}}}},{"id":"ec-m06a-nourse-1955-automation-hearings-statement-a","statement":"Nourse, “Statement (former Chairman, Council of Economic Advisers), in Automation and Technological Change: Hearings” (1955-10-28) articulates M06a. Speaker: Edwin G. Nourse, former Chairman, Council of Economic Advisers (economist). Nourse asks whether automation will disturb the price-and-income relations behind the postwar prosperity, and lists its effects on wages, capital, costs, prices, profits and taxes.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Applies to some groups and places, not aggregate employment. The quote stops before a line-break hyphen ('down¬graded'). Matches the cached .txt.","evidenceIds":["evidence-ec-m06a-03"],"attributes":{"thought":{"roles":["articulation"],"register":"nonfiction","tier":{"value":1,"basis":"Prepared statement by an economist (former Chairman of the Council of Economic Advisers, shown in the text); a substantive argued statement. Tier 1 by maintainer decision of 2026-10-12 (economist testimony)."},"conditions":"Mixed effects: the sentence continues that wage rates will be 'upgraded here and downgraded there'."}}},{"id":"lg-oai-brockman-president-2022","statement":"OpenAI announced on 5 May 2022 that Greg Brockman was becoming President, a new role.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The announcement does not give an exact start date or establish a continuing role.","evidenceIds":["lg-ev-oai-brockman-president-2022"],"attributes":{"role":"President"}},{"id":"lg-cn-01ai-founded-2023","statement":"01.AI’s company page states that the company was established in May 2023.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"A retrospective statement on a mutable company page, not a registration record. English rendering of the Chinese passage is a machine translation; no human translation review.","evidenceIds":["lg-ev-cn-01ai-founded-2023"]},{"id":"lg-mx-wu-xai-team-2023","statement":"Ars Technica reported on 12 July 2023 that xAI's website listed Tony Wu among the members of its team at launch.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The record rests on reporting of xAI's launch web page, which was not retrieved directly. The source names 'Tony Wu'; identity with Yuhuai (Tony) Wu rests on his later self-identification as departing xAI.","evidenceIds":["lg-ev-mx-wu-xai-team-2023"],"attributes":{"role":"Launch-team member"}},{"id":"lg-ant-krieger-cpo-2024","statement":"Anthropic announced on 15 May 2024 that Mike Krieger had joined it as Chief Product Officer.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Anthropic's January 2026 Labs announcement describes him as having spent the past two years as Chief Product Officer and as joining Labs; no exact end date of the CPO role is stated.","evidenceIds":["lg-ev-ant-krieger-cpo-2024","lg-ev-ant-krieger-cpo-2024-labs"],"attributes":{"role":"Chief Product Officer"}},{"id":"lg-gw-weng-openai-return-2026","statement":"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.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"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.","evidenceIds":["lg-ev-gw-weng-openai-return-2026-a","lg-ev-gw-weng-openai-return-2026-b"],"attributes":{"activityStage":"announced","role":"lead of a top-level team focused on accelerating internal research"}},{"id":"lg-gs-shah-gdm-asat-lead","statement":"Rohin Shah's homepage, as retrieved on 15 October 2026, states that he leads the AGI Safety & Alignment team at Google DeepMind.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Self-description on an undated page; an undated page can lag behind changes. It does not establish when he began to lead the team.","evidenceIds":["lg-ev-gs-shah-gdm-asat-lead"],"attributes":{"role":"leads the AGI Safety & Alignment team"}},{"id":"lg-gw-antonoglou-cofounded-reflection","statement":"Sequoia Capital’s announcement of 7 March 2025 names Ioannis Antonoglou as a co-founder of Reflection AI.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The record rests on an investor’s announcement of its partnership with the company; no Reflection AI page naming its founders was retrieved. It does not establish the founding date or the completeness of the founder roster.","evidenceIds":["lg-ev-gw-antonoglou-cofounded-reflection"]},{"id":"lg-gw-kantor-cofounded-h-company","statement":"TechCrunch reported on 21 May 2024 that Charles Kantor was H’s co-founder.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The record rests on reporting (TechCrunch); no H Company page naming its founders was retrieved. It does not establish the founding date. The same sentence calls him CEO in May 2024; H’s about page states that Gautier Cloix became CEO in 2025.","evidenceIds":["lg-ev-gw-kantor-cofounded-h-company","lg-ev-gw-kantor-cofounded-h-company-cloix"],"attributes":{"role":"co-founder"}},{"id":"cmp-rsp-v2-1-d6-competitor","statement":"RSP · v2.1 states a provision on the comparison question “Provisions conditioned on other developers”.","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, and a shared question does not make two editions’ provisions equivalent.","evidenceIds":["ev-cmp-rsp-v2-1-d6-competitor-1"],"attributes":{"provision":{"kind":"exception","key":"comparison-d6-competitor","facet":{"id":"D6.competitor","status":"stated"}}}},{"id":"cmp-rsp-v2-1-d6-discretion","statement":"RSP · v2.1 states a provision on the comparison question “Other exceptions or discretion”.","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, and a shared question does not make two editions’ provisions equivalent.","evidenceIds":["ev-cmp-rsp-v2-1-d6-discretion-1"],"attributes":{"provision":{"kind":"exception","key":"comparison-d6-discretion","facet":{"id":"D6.discretion","status":"stated"}}}},{"id":"ap-ajeya-cotra-cotra-2022-without-specific-countermeasures","statement":"“Without specific countermeasures, the easiest path to transformative AI likely leads to AI takeover” (2022-07-18) credits Ajeya Cotra 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-ajeya-cotra-cotra-2022-without-specific-countermeasures"]},{"id":"ap-vladimir-mikulik-rlo-2019-introduction","statement":"“Risks from Learned Optimization: Introduction” (2019-05-31) credits Vlad Mikulik as one of its 5 authors.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Credited co-authorship; it does not allocate parts of the text between authors or imply agreement with every passage.","evidenceIds":["evidence-ap-vladimir-mikulik-rlo-2019-introduction"]},{"id":"ap-eliezer-yudkowsky-lw-2008-ghosts-in-the-machine","statement":"“Ghosts in the Machine” (2008-06-17) credits Eliezer Yudkowsky 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-eliezer-yudkowsky-lw-2008-ghosts-in-the-machine"]},{"id":"cmp-rsp-v2-1-d1-cbrn","statement":"RSP · v2.1 states a provision on the comparison question “Chemical, biological, radiological or nuclear weapons”.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"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-rsp-v2-1-d1-cbrn-1"],"attributes":{"provision":{"kind":"risk_domain","key":"comparison-d1-cbrn","facet":{"id":"D1.cbrn","status":"stated"}}}},{"id":"cmp-rsp-v2-1-d7-external","statement":"RSP · v2.1 states a provision on the comparison question “Public or external reporting”.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"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-rsp-v2-1-d7-external-1"],"attributes":{"provision":{"kind":"reporting","key":"comparison-d7-external","facet":{"id":"D7.external","status":"stated"}}}},{"id":"cmp-fsf-v2-0-d2-levels","statement":"FSF · v2.0 states a provision on the comparison question “Named levels or thresholds”.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Misuse CCLs (Table 1): CBRN uplift 1, Cyber autonomy level 1, Cyber uplift level 1, ML R&D uplift level 1, ML R&D autonomy level 1. Deceptive alignment CCLs (Table 2): Instrumental Reasoning Levels 1 and 2. Early warning evaluations carry an “alert threshold” that signals that a CCL could be reached ahead of the next evaluation round (PDF p. 3). 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-v2-0-d2-levels-1"],"attributes":{"provision":{"kind":"threshold","key":"comparison-d2-levels","facet":{"id":"D2.levels","status":"stated"}}}},{"id":"cmp-frontier-governance-2026-d8-process","statement":"FGF · May 2026 edition states a provision on the comparison question “Revision process and approval”.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Being presented for oversight is not stated as approval: the edition names no approver. §7.1 states that updates may be proposed by named officers and bodies, that the Legal function oversees the update process, and that changelogs for material updates are published within 30 days (PDF p. 21). The edition describes itself as overlapping with the Preparedness Framework, which OpenAI says it will keep using and developing (PDF p. 4). 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-frontier-governance-2026-d8-process-1","ev-cmp-frontier-governance-2026-d8-process-2"],"attributes":{"provision":{"kind":"revision","key":"comparison-d8-process","facet":{"id":"D8.process","status":"stated"}}}},{"id":"cmp-faif-2025-d5-cadence","statement":"FAIF · December 2025 edition states a provision on the comparison question “Evaluation timing”.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"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-2025-d5-cadence-1"],"attributes":{"provision":{"kind":"evaluation_timing","key":"comparison-d5-cadence","facet":{"id":"D5.cadence","status":"stated"}}}},{"id":"c2-baidu-20f-fy2025-ernie-5","statement":"Baidu's annual report on Form 20-F, filed on 17 March 2026, states that Baidu unveiled ERNIE 5.0 at Baidu World 2025 in November 2025 and released an updated version of ERNIE 5.0 in January 2026.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Baidu's own account in a securities filing. The filing's descriptions of ERNIE 5.0's capabilities are not recorded. No Baidu announcement of the November 2025 event was retrieved for this pack.","evidenceIds":["c2-ev-baidu-20f-fy2025-ernie-5","c2-ev-baidu-20f-fy2025-ernie-5-update"]},{"id":"ap-katja-grace-grace-2014-superintelligence-reading-group","statement":"“Superintelligence reading group” (2014-08-31) credits the username “KatjaGrace” as its author.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Credited under the username “KatjaGrace”. The account’s post “Atari early” (LessWrong, 2020-04-02) carries the same username byline and, in its body, the author line “By Katja Grace”, which identifies the account. It says nothing about priority for the ideas discussed.","evidenceIds":["evidence-ap-katja-grace-grace-2014-superintelligence-reading-group"]},{"id":"fl-link-rsp-v3-1-sb53","statement":"Anthropic’s Responsible Scaling Policy v3.1 keeps the footnote saying that where laws such as California SB 53 define “catastrophic risk” or similar terms with specific thresholds, Anthropic addresses those requirements in separate compliance frameworks.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The edition’s own statement about SB 53, quoted from it. It names SB 53 and points to separate compliance frameworks; it does not say that the RSP is a framework under the Act. It is not evidence that the edition satisfies the statute, that its publisher is a frontier developer under it, or of implementation or compliance. The record on v3.0 notes the same footnote in the later editions; this record gives v3.1 its own link, as the dataset does for the White House and Seoul links.","evidenceIds":["evidence-fl-link-rsp-v3-1-sb53-1"],"attributes":{"commitment":{"link":"names"}}},{"id":"fl-intl-bletchley-developer-responsibility","statement":"The Bletchley Declaration affirms that actors developing frontier AI capabilities, in particular unusually powerful and potentially harmful systems, have a 'particularly strong responsibility' for ensuring their safety, including through safety testing and evaluations.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"An affirmation by the participating governments; it imposes no legal duty and names no developer.","evidenceIds":["evidence-fl-intl-bletchley-developer-responsibility-1"],"attributes":{"provision":{"kind":"required_action","key":"intl-bletchley-developer-responsibility"}}},{"id":"ab-scr-ostp-framework-attestation","statement":"The Framework states that to adhere to it a provider or manufacturer should either post on a public webpage a statement self-attesting to following each of its criteria or give such a statement to the federally funded customer and the funding agency on request, and that the attestation should be signed by someone authorised for the organisation, give a point of contact, be updated by 1 January each year, and commit the organisation to notify customers and funding agencies within 72 hours if it ceases to adhere.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"States what the framework asks of a provider or manufacturer. A self-attestation is a statement by the organisation making it; it is not verification, certification or compliance, and no record is made of any organisation having attested.","evidenceIds":["evidence-ab-scr-ostp-framework-attestation-1","evidence-ab-scr-ostp-framework-attestation-2"],"attributes":{"provision":{"kind":"required_action","key":"us-ostp-nucleic-acid-framework-attestation"}}},{"id":"ab-gov-us-eo14110-sec41b-doe-tools","statement":"Section 4.1(b) of Executive Order 14110 directs the Secretary of Energy, within 270 days, to develop and, to the extent permitted by law and available appropriations, implement a plan for developing the Department of Energy's AI model evaluation tools and AI testbeds, and provides that the Secretary shall at a minimum develop tools to evaluate AI capabilities to generate outputs that may represent nuclear, nonproliferation, biological, chemical, critical infrastructure and energy-security threats or hazards, doing this work solely to guard against those threats and also developing model guardrails that reduce such risks.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"A directive to develop a plan, tools and guardrails. It is not evidence that any plan, tool, testbed or guardrail was made or works, and it states no evaluation result. The subsection also directs consultation with private AI laboratories, academia, civil society and third-party evaluators.","evidenceIds":["evidence-ab-gov-us-eo14110-sec41b-doe-tools-1","evidence-ab-gov-us-eo14110-sec41b-doe-tools-2"],"attributes":{"provision":{"kind":"required_action","key":"us-eo14110-sec41b-doe-evaluation-tools"}}},{"id":"ab-ev-vct-benchmark-measures","statement":"The Virology Capabilities Test paper states that it presents a large language model benchmark measuring the capability to troubleshoot complex virology laboratory protocols, built from the inputs of dozens of expert virologists.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Records what the benchmark states it measures, as described in its abstract, not any finding about a system. Only the abstract page was read; the paper's construction and contents were not inspected. The quoted phrase is the benchmark's own statement of what it measures and is required to identify it; no question, protocol or procedure is recorded, and the paper body was deliberately not read.","evidenceIds":["evidence-ab-ev-vct-benchmark-measures-1"]},{"id":"ab-arg-d2-brent-contests","statement":"The RAND working paper of 12 June 2025 states that safety assessments mistakenly assume that biological weapons development requires tacit knowledge, that is, expertise gained through experience that cannot be conveyed in words.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The record documents what this working paper states about an assumption it attributes to other assessments; it is not evidence that the proposition is false, and the paper is marked by its publisher as not formally reviewed or edited. The paper's case studies and dialogues contain technical content. Only the summary's statement about the assumption is cited; that content was excluded and is not paraphrased.","evidenceIds":["evidence-ab-arg-d2-brent-contests-1","evidence-ab-arg-d2-brent-contests-2"],"attributes":{"thought":{"roles":["contesting"],"register":"nonfiction","qualifier":"Contests the premise as the paper states other assessments hold it.","engagement":{"kind":"engages-premise","rationale":"The paper names the tacit-knowledge premise and argues against it from historical cases."}}}}],"sources":[{"id":"source-good","title":"Speculations Concerning the First Ultraintelligent Machine","authors":["I. J. Good"],"publisher":"Academic Press","publicationDate":"1965","version":"Scanned reprint, Advances in Computers volume 6, pp. 31–88. Repository metadata identifies ©1965; the mirror filename says 1966.","canonicalUrl":"https://gwern.net/doc/ai/1966-good.pdf","evidenceUrl":"https://gwern.net/doc/ai/1966-good.pdf","language":"en","accessedAt":"2026-09-17","reuse":{"attribution":"Academic Press, Speculations Concerning the First Ultraintelligent Machine (1965).","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and minimal attributed excerpts only. Recheck terms before wider redistribution."}},{"id":"source-good-catalogue","title":"Repository record: Speculations Concerning the First Ultraintelligent Machine","authors":["Virginia Tech"],"publisher":"Virginia Tech","publicationDate":"2005-03-05","version":"Metadata for technical report 05-3, issued 2005 and accessioned 2019. The notes identify the original as a 1965 reprint.","canonicalUrl":"https://vtechworks.lib.vt.edu/items/5085379d-b24c-424e-8861-e70a47b4b2fb/full","evidenceUrl":"https://vtechworks.lib.vt.edu/items/5085379d-b24c-424e-8861-e70a47b4b2fb/full","language":"en","archivedUrl":"https://web.archive.org/web/20260419044157/https://vtechworks.lib.vt.edu/items/5085379d-b24c-424e-8861-e70a47b4b2fb/full","archivedAt":"2026-04-19","accessedAt":"2026-09-17","reuse":{"attribution":"Virginia Tech, repository metadata for Technical Report No. 05-3.","licence":"In Copyright (as stated in the repository record)","note":"Bibliographic metadata only; no full-text redistribution."}},{"id":"source-anthropic-reinvent-2024","title":"Leverage Anthropic’s Claude models for AI’s evolving landscape","authors":["Maggie Vo","Elie Schoppik"],"publisher":"Anthropic / AWS re:Invent","publicationDate":"2024","version":"AIM123-S; 2024 deck as identified by copyright and event path. PDF page 6 contains founder rows. No exact presentation or upload date inferred.","canonicalUrl":"https://d1.awsstatic.com/onedam/marketing-channels/website/aws/en_US/events/approved/reinvent-2025/reinvent/2024/slides/aim/AIM123-S_Leverage-Anthropics-Claude-models-for-AIs-evolving-landscape-sponsored-by-Anthropic.pdf","evidenceUrl":"https://d1.awsstatic.com/onedam/marketing-channels/website/aws/en_US/events/approved/reinvent-2025/reinvent/2024/slides/aim/AIM123-S_Leverage-Anthropics-Claude-models-for-AIs-evolving-landscape-sponsored-by-Anthropic.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20261029121322/https://d1.awsstatic.com/onedam/marketing-channels/website/aws/en_US/events/approved/reinvent-2025/reinvent/2024/slides/aim/AIM123-S_Leverage-Anthropics-Claude-models-for-AIs-evolving-landscape-sponsored-by-Anthropic.pdf","archivedAt":"2026-10-29","accessedAt":"2026-09-18","reuse":{"attribution":"Maggie Vo; Elie Schoppik, Leverage Anthropic’s Claude models for AI’s evolving landscape (2024).","licence":"No open redistribution licence asserted","note":"Bibliographic metadata and brief attributed paraphrases or excerpts only. Source roles are relative to each claim."}},{"id":"source-xai-team-2023","title":"Musk announces new AI company that seeks to “understand the universe”","authors":["Benj Edwards"],"publisher":"Ars Technica","publicationDate":"2023-07-12","version":"Contemporaneous report and screenshot of the launch roster. The current x.ai company page has changed; original archive retrieval was unavailable.","canonicalUrl":"https://arstechnica.com/information-technology/2023/07/musk-announces-new-ai-company-that-seeks-to-understand-the-universe/","evidenceUrl":"https://arstechnica.com/information-technology/2023/07/musk-announces-new-ai-company-that-seeks-to-understand-the-universe/","language":"en","archivedUrl":"https://web.archive.org/web/20260211075334/https://arstechnica.com/information-technology/2023/07/musk-announces-new-ai-company-that-seeks-to-understand-the-universe/","archivedAt":"2026-02-11","accessedAt":"2026-09-18","reuse":{"attribution":"Benj Edwards, Musk announces new AI company that seeks to “understand the universe” (2023-07-12).","licence":"No open redistribution licence asserted","note":"Bibliographic metadata and brief attributed paraphrases or excerpts only. Source roles are relative to each claim."}},{"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-3-1","title":"Responsible Scaling Policy — v3.1","authors":["Anthropic"],"publisher":"Anthropic","publicationDate":null,"version":"Responsible Scaling Policy v3.1; effective date 2026-04-02.","versionDate":"2026-04-02","canonicalUrl":"https://www-cdn.anthropic.com/files/4zrzovbb/website/bf04581e4f329735fd90634f6a1962c13c0bd351.pdf","evidenceUrl":"https://www-cdn.anthropic.com/files/4zrzovbb/website/bf04581e4f329735fd90634f6a1962c13c0bd351.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20260407193305/https://www-cdn.anthropic.com/files/4zrzovbb/website/bf04581e4f329735fd90634f6a1962c13c0bd351.pdf","archivedAt":"2026-04-07","accessedAt":"2026-09-18","reuse":{"attribution":"Anthropic, Responsible Scaling Policy — 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-fsf-2-0","title":"Frontier Safety Framework — v2.0","authors":["Google DeepMind"],"publisher":"Google DeepMind","publicationDate":"2025-02-04","version":"Frontier Safety Framework v2.0; publication date 2025-02-04.","versionDate":"2025-02-04","canonicalUrl":"https://storage.googleapis.com/deepmind-media/DeepMind.com/Blog/updating-the-frontier-safety-framework/Frontier%20Safety%20Framework%202.0.pdf","evidenceUrl":"https://storage.googleapis.com/deepmind-media/DeepMind.com/Blog/updating-the-frontier-safety-framework/Frontier%20Safety%20Framework%202.0.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20250205011727/https://storage.googleapis.com/deepmind-media/DeepMind.com/Blog/updating-the-frontier-safety-framework/Frontier%20Safety%20Framework%202.0.pdf","archivedAt":"2025-02-05","accessedAt":"2026-09-18","reuse":{"attribution":"Google DeepMind, Frontier Safety Framework — v2.0.","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-fgf-2026","title":"Frontier Governance Framework — May 2026 edition","authors":["OpenAI"],"publisher":"OpenAI","publicationDate":"2026-05-28","version":"Frontier Governance Framework May 2026 edition; publication date 2026-05-28.","versionDate":"2026-05-28","canonicalUrl":"https://cdn.openai.com/pdf/e37d949b-8c9f-4d76-b99e-4272f4631a7e/openai-frontier-governance-framework.pdf","evidenceUrl":"https://cdn.openai.com/pdf/e37d949b-8c9f-4d76-b99e-4272f4631a7e/openai-frontier-governance-framework.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20260529171758/https://cdn.openai.com/pdf/e37d949b-8c9f-4d76-b99e-4272f4631a7e/openai-frontier-governance-framework.pdf","archivedAt":"2026-05-29","accessedAt":"2026-09-18","reuse":{"attribution":"OpenAI, Frontier Governance Framework — May 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-faif-2025","title":"Frontier Artificial Intelligence Framework — December 2025 edition","authors":["xAI"],"publisher":"xAI","publicationDate":null,"version":"Frontier Artificial Intelligence Framework December 2025 edition; revision date 2025-12-30.","versionDate":"2025-12-30","canonicalUrl":"https://data.x.ai/2025-12-31-xai-frontier-artificial-intelligence-framework.pdf","evidenceUrl":"https://data.x.ai/2025-12-31-xai-frontier-artificial-intelligence-framework.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20260102171656/https://data.x.ai/2025-12-31-xai-frontier-artificial-intelligence-framework.pdf","archivedAt":"2026-01-02","accessedAt":"2026-09-18","reuse":{"attribution":"xAI, Frontier Artificial Intelligence Framework — December 2025 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-gemini-launch-2023","title":"Introducing Gemini: Google’s most capable AI model yet","authors":["Google"],"publisher":"Google","publicationDate":"2023-12-06","version":"Launch article by Sundar Pichai and Demis Hassabis on behalf of the Gemini team; displayed article title is now Introducing Gemini: Google’s most capable AI model yet. December 13 update note concerns messaging apps.","canonicalUrl":"https://blog.google/innovation-and-ai/technology/ai/google-gemini-ai/","evidenceUrl":"https://blog.google/innovation-and-ai/technology/ai/google-gemini-ai/","language":"en","archivedUrl":"https://web.archive.org/web/20260914020056/https://blog.google/innovation-and-ai/technology/ai/google-gemini-ai/","archivedAt":"2026-09-14","accessedAt":"2026-09-19","reuse":{"attribution":"Google, Introducing Gemini: our largest and most capable AI model (2023-12-06).","licence":"No open redistribution licence asserted","note":"Bibliographic metadata and editorial paraphrases only; no full source redistributed."}},{"id":"source-atlas-deepmind-feedback","title":"Learning through human feedback","authors":["Jan Leike","Miljan Martic","Shane Legg"],"publisher":"DeepMind","publicationDate":"2017-06-12","version":"Dated publisher page inspected 2026-09-28; current HTML, not an authenticated historical capture.","canonicalUrl":"https://deepmind.google/blog/learning-through-human-feedback/","evidenceUrl":"https://deepmind.google/blog/learning-through-human-feedback/","language":"en","archivedUrl":"https://web.archive.org/web/20260925235348/https://deepmind.google/blog/learning-through-human-feedback/","archivedAt":"2026-09-25","accessedAt":"2026-09-28","reuse":{"attribution":"Jan Leike; Miljan Martic; Shane Legg. Learning through human feedback.","licence":"No redistribution licence asserted.","note":"Short attributed paraphrases only; original publication is linked, not redistributed."}},{"id":"source-ms-fgf-2025","title":"Frontier Governance Framework — Version 1 (February 2025)","authors":["Microsoft"],"publisher":"Microsoft","publicationDate":"2025-02","version":"Frontier Governance Framework, Version 1 (February 2025); its change log reads “8 February 2025 – First version”. The file served at the recorded address (last modified 2025-02-12) is byte-identical to the Internet Archive capture of 2025-02-26, the earliest located.","versionDate":"2025-02-08","canonicalUrl":"https://cdn-dynmedia-1.microsoft.com/is/content/microsoftcorp/microsoft/final/en-us/microsoft-brand/documents/Microsoft-Frontier-Governance-Framework.pdf","evidenceUrl":"https://cdn-dynmedia-1.microsoft.com/is/content/microsoftcorp/microsoft/final/en-us/microsoft-brand/documents/Microsoft-Frontier-Governance-Framework.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20250226161803/https://cdn-dynmedia-1.microsoft.com/is/content/microsoftcorp/microsoft/final/en-us/microsoft-brand/documents/Microsoft-Frontier-Governance-Framework.pdf","archivedAt":"2025-02-26","accessedAt":"2026-10-03","reuse":{"attribution":"Microsoft, Frontier Governance Framework, Version 1 (February 2025).","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-wiener-1960-science","title":"Some Moral and Technical Consequences of Automation","authors":["Norbert Wiener"],"publisher":"Science (American Association for the Advancement of Science)","publicationDate":"1960-05-06","version":"Scan of the original Science pages in a later JSTOR download, hosted on a Pennsylvania State University course page; text layer read and pp. 1357–1358 compared with the rendered pages by the pre-1996 candidate search.","canonicalUrl":"https://doi.org/10.1126/science.131.3410.1355","evidenceUrl":"https://faculty.ist.psu.edu/vhonavar/Courses/ai100/Wiener.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20241222023645/https://faculty.ist.psu.edu/vhonavar/Courses/ai100/Wiener.pdf","archivedAt":"2024-12-22","accessedAt":"2026-10-06","reuse":{"attribution":"Norbert Wiener, \"Some Moral and Technical Consequences of Automation\", Science 131 (1960), 1355–1358.","licence":"No open redistribution licence asserted; the inspected copy is a university course file.","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"source-extropians-1996-08-30-clemmensen","title":"Re: Thinking about the future... (Extropians mailing list)","authors":["Dan Clemmensen"],"publisher":"Extropians mailing list","publicationDate":"1996-08-30","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.","canonicalUrl":"http://extropians.weidai.com/extropians.96/0613.html","evidenceUrl":"http://extropians.weidai.com/extropians.96/0613.html","language":"en","archivedUrl":"https://web.archive.org/web/20150427133328/http://extropians.weidai.com/extropians.96/0613.html","archivedAt":"2015-04-27","accessedAt":"2026-10-06","reuse":{"attribution":"Dan Clemmensen, \"Re: Thinking about the future...\", Extropians mailing list, 1996-08-30.","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":"source-extropians-1996-09-07-clemmensen","title":"Re: SI ultimatum speculation (Extropians mailing list)","authors":["Dan Clemmensen"],"publisher":"Extropians mailing list","publicationDate":"1996-09-07","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.","canonicalUrl":"http://extropians.weidai.com/extropians.96/0841.html","evidenceUrl":"http://extropians.weidai.com/extropians.96/0841.html","language":"en","archivedUrl":"https://web.archive.org/web/20150427201436/http://extropians.weidai.com/extropians.96/0841.html","archivedAt":"2015-04-27","accessedAt":"2026-10-06","reuse":{"attribution":"Dan Clemmensen, \"Re: SI ultimatum speculation\", Extropians mailing list, 1996-09-07.","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-sits-2000-capture","title":"Staring into the Singularity (version 1.2.1)","authors":["Eliezer S. Yudkowsky"],"publisher":"Eliezer S. Yudkowsky","publicationDate":"1996-11-18","version":"Version 1.2.1, \"Updated: 08/31/2000\", in the raw Internet Archive capture of 2000-12-04; retrieved 2026-10-06 with the hash recorded by the 1996–2022 candidate search. Personal website, sysopmind.com.","versionDate":"2000-08-31","canonicalUrl":"http://sysopmind.com/singularity.html","evidenceUrl":"https://web.archive.org/web/20001204213100/http://sysopmind.com:80/singularity.html","language":"en","archivedUrl":"https://web.archive.org/web/20001204213100/http://sysopmind.com:80/singularity.html","archivedAt":"2000-12-04","accessedAt":"2026-10-06","reuse":{"attribution":"Eliezer S. Yudkowsky, \"Staring into the Singularity\", version 1.2.1 (2000).","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-extropians-1996-12-31-yudkowsky-5142","title":"Re: Goal-based AI (Extropians mailing list)","authors":["Eliezer Yudkowsky"],"publisher":"Extropians mailing list","publicationDate":"1996-12-31","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.","canonicalUrl":"http://extropians.weidai.com/extropians.96/5142.html","evidenceUrl":"http://extropians.weidai.com/extropians.96/5142.html","language":"en","archivedUrl":"https://web.archive.org/web/20160414031350/http://extropians.weidai.com/extropians.96/5142.html","archivedAt":"2016-04-14","accessedAt":"2026-10-06","reuse":{"attribution":"Eliezer Yudkowsky, \"Re: Goal-based AI\", Extropians mailing list, 1996-12-31.","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-grace-2014-superintelligence-reading-group","title":"Superintelligence reading group","authors":["Katja Grace"],"publisher":"LessWrong","publicationDate":"2014-08-31","version":"Text checked in the post’s first stored version through the LessWrong GraphQL API on 2026-10-06; content version 1.0.0, never edited.","canonicalUrl":"https://www.lesswrong.com/posts/QDmzDZ9CEHrKQdvcn/superintelligence-reading-group","evidenceUrl":"https://www.lesswrong.com/posts/QDmzDZ9CEHrKQdvcn/superintelligence-reading-group","language":"en","archivedUrl":"https://web.archive.org/web/20260917115136/https://www.lesswrong.com/posts/QDmzDZ9CEHrKQdvcn/superintelligence-reading-group","archivedAt":"2026-09-17","accessedAt":"2026-10-06","reuse":{"attribution":"Katja Grace, \"Superintelligence reading group\", LessWrong, 2014-08-31.","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-christiano-2018-takeoff-speeds","title":"Takeoff speeds","authors":["Paul Christiano"],"publisher":"The sideways view (Paul Christiano’s blog)","publicationDate":"2018-02-24","version":"Current page, retrieved 2026-10-06; its bytes differ from the 1996–2022 candidate search’s retrieval while the quoted text matches. A machine check compared the capture of 2018-02-25, which was retrieved again on 2026-10-06.","canonicalUrl":"https://sideways-view.com/2018/02/24/takeoff-speeds/","evidenceUrl":"https://sideways-view.com/2018/02/24/takeoff-speeds/","language":"en","archivedUrl":"https://web.archive.org/web/20180225225416/https://sideways-view.com/2018/02/24/takeoff-speeds/","archivedAt":"2018-02-25","accessedAt":"2026-10-06","reuse":{"attribution":"Paul Christiano, \"Takeoff speeds\", The sideways view, 24 February 2018.","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-rlo-2019-introduction","title":"Risks from Learned Optimization: Introduction","authors":["Evan Hubinger","Chris van Merwijk","Vladimir Mikulik","Joar Skalse","Scott Garrabrant"],"publisher":"Alignment Forum and LessWrong","publicationDate":"2019-05-31","version":"Text checked in the post’s first stored version through the LessWrong GraphQL API on 2026-10-06; content version 1.0.0 checked; the current version is later.","canonicalUrl":"https://www.alignmentforum.org/posts/FkgsxrGf3QxhfLWHG/risks-from-learned-optimization-introduction","evidenceUrl":"https://www.alignmentforum.org/posts/FkgsxrGf3QxhfLWHG/risks-from-learned-optimization-introduction","language":"en","archivedUrl":"https://web.archive.org/web/20260714195403/https://www.alignmentforum.org/posts/FkgsxrGf3QxhfLWHG/risks-from-learned-optimization-introduction","archivedAt":"2026-07-14","accessedAt":"2026-10-06","reuse":{"attribution":"Evan Hubinger, Chris van Merwijk, Vladimir Mikulik, Joar Skalse, Scott Garrabrant, \"Risks from Learned Optimization: Introduction\", Alignment Forum and LessWrong, 2019-05-31.","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-cotra-2022-without-specific-countermeasures","title":"Without specific countermeasures, the easiest path to transformative AI likely leads to AI takeover","authors":["Ajeya Cotra"],"publisher":"Alignment Forum and LessWrong","publicationDate":"2022-07-18","version":"Text checked in the post’s first stored version through the LessWrong GraphQL API on 2026-10-06; content version 1.0.0.","canonicalUrl":"https://www.lesswrong.com/posts/pRkFkzwKZ2zfa3R6H/without-specific-countermeasures-the-easiest-path-to","evidenceUrl":"https://www.lesswrong.com/posts/pRkFkzwKZ2zfa3R6H/without-specific-countermeasures-the-easiest-path-to","language":"en","archivedUrl":"https://web.archive.org/web/20220718200028/https://www.lesswrong.com/posts/pRkFkzwKZ2zfa3R6H/without-specific-countermeasures-the-easiest-path-to","archivedAt":"2022-07-18","accessedAt":"2026-10-06","reuse":{"attribution":"Ajeya Cotra, \"Without specific countermeasures, the easiest path to transformative AI likely leads to AI takeover\", Alignment Forum and LessWrong, 2022-07-18.","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-lw-2008-ghosts-in-the-machine","title":"Ghosts in the Machine","authors":["Eliezer Yudkowsky"],"publisher":"Overcoming Bias","publicationDate":"2008-06-17","version":"Post body retrieved through the LessWrong GraphQL API on 2026-10-06 (SHA-256 e86e1702da19cf0155ee1218fd7ac85d8598e0dca6b2d2b1c8738c669b83fa65), content version 1.0.0. As imported to LessWrong.","canonicalUrl":"https://www.lesswrong.com/posts/cnYHFNBF3kZEyx24v/ghosts-in-the-machine","evidenceUrl":"https://www.lesswrong.com/posts/cnYHFNBF3kZEyx24v/ghosts-in-the-machine","language":"en","archivedUrl":"https://web.archive.org/web/20260920013935/https://www.lesswrong.com/posts/cnYHFNBF3kZEyx24v/ghosts-in-the-machine","archivedAt":"2026-09-20","accessedAt":"2026-10-06","reuse":{"attribution":"Eliezer Yudkowsky, “Ghosts in the Machine”, Overcoming Bias, as imported to LessWrong, 2008-06-17.","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-sl4-2000-12-10-yudkowsky-0012-0355","title":"Revising a Friendly AI (SL4 mailing list)","authors":["Eliezer Yudkowsky"],"publisher":"SL4 mailing list","publicationDate":"2000-12-10","version":"Archive HTML as served, retrieved 2026-10-07 (SHA-256 4fb349f96304ea9c661f22830b6831e3a8dba94ea90584ebabcf9a4ff7e46263); 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/0012/0355.html","evidenceUrl":"http://sl4.org/archive/0012/0355.html","language":"en","archivedUrl":"https://web.archive.org/web/20260408020603/http://sl4.org/archive/0012/0355.html","archivedAt":"2026-04-08","accessedAt":"2026-10-07","reuse":{"attribution":"Eliezer Yudkowsky, “Revising a Friendly AI”, SL4 mailing list, 2000-12-10.","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-sl4-2005-07-13-yudkowsky-0507-11483","title":"Re: Fighting UFAI (SL4 mailing list)","authors":["Eliezer Yudkowsky"],"publisher":"SL4 mailing list","publicationDate":"2005-07-13","version":"Archive HTML as served, retrieved 2026-10-07 (SHA-256 7c6b98d1ab322db0e7bdbcc1cb9297921bb54f76781dae7865eaf188711aab56); 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/0507/11483.html","evidenceUrl":"http://sl4.org/archive/0507/11483.html","language":"en","archivedUrl":"https://web.archive.org/web/20251109194726/http://sl4.org/archive/0507/11483.html","archivedAt":"2025-11-09","accessedAt":"2026-10-07","reuse":{"attribution":"Eliezer Yudkowsky, “Re: Fighting UFAI”, SL4 mailing list, 2005-07-13.","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-nourse-1955-automation-hearings-statement","title":"Statement (former Chairman, Council of Economic Advisers), in Automation and Technological Change: Hearings","authors":["Edwin G. Nourse"],"publisher":"U.S. Congress, Joint Committee on the Economic Report, Subcommittee on Economic Stabilization. (1955). Automation and technological change: Hearings … 84th Congress, 1st session, October 14–28, 1955. Washington: U.S. Government Printing Office, from p. 618","publicationDate":"1955-10-28","version":"Internet Archive OCR text (djvu.txt) of a scanned copy, retrieved 2026-10-09 from https://ia803102.us.archive.org/32/items/congress-automation-and-technological-change/congress-automation-and-technological-changedjvu.txt; OCR errors possible; quotes checked only against clean stretches. Current copy; no earlier capture compared. SHA-256 3962809fbb923be62bf986addb0014ee147934a8767475c1f7cfe051959bff90.","canonicalUrl":"https://archive.org/details/congress-automation-and-technological-change","evidenceUrl":"https://archive.org/details/congress-automation-and-technological-change","language":"en","accessedAt":"2026-10-09","reuse":{"attribution":"Edwin G. Nourse, “Statement (former Chairman, Council of Economic Advisers), in Automation and Technological Change: Hearings”, U.S. Congress, Joint Committee on the Economic Report, Subcommittee on Economic Stabilization. (1955). Automation and technological change: Hearings … 84th Congress, 1st session, October 14–28, 1955. Washington: U.S. Government Printing Office, from p. 618, 1955-10-28.","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-ant-krieger-2024","title":"Mike Krieger joins Anthropic as Chief Product Officer","authors":["Anthropic"],"publisher":"Anthropic","publicationDate":"2024-05-15","version":"Dated page, retrieved on 2026-10-15 (SHA-256 3588e9ddfc7a37c697a82c03f3ba26e730ba76b130aa769d45e8306ebc8bea86) directly from the publisher.","canonicalUrl":"https://www.anthropic.com/news/mike-krieger-joins-anthropic","evidenceUrl":"https://www.anthropic.com/news/mike-krieger-joins-anthropic","language":"en","archivedUrl":"https://web.archive.org/web/20260926220542/https://www.anthropic.com/news/mike-krieger-joins-anthropic","archivedAt":"2026-09-26","accessedAt":"2026-10-15","reuse":{"attribution":"Anthropic, “Mike Krieger joins Anthropic as Chief Product Officer”, 2024-05-15.","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-ant-labs-2026","title":"Introducing Labs","authors":["Anthropic"],"publisher":"Anthropic","publicationDate":"2026-01-13","version":"Dated page, retrieved on 2026-10-15 (SHA-256 a264ec6e85bdb01d1485b31234e2ab077f740576e2c8cafc9c5aad34b25a463a) directly from the publisher.","canonicalUrl":"https://www.anthropic.com/news/introducing-anthropic-labs","evidenceUrl":"https://www.anthropic.com/news/introducing-anthropic-labs","language":"en","archivedUrl":"https://web.archive.org/web/20260926162531/https://www.anthropic.com/news/introducing-anthropic-labs","archivedAt":"2026-09-26","accessedAt":"2026-10-15","reuse":{"attribution":"Anthropic, “Introducing Labs”, 2026-01-13.","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-cn-01ai-about","title":"关于我们 — 01.AI 零一万物","authors":["01.AI"],"publisher":"01.AI (零一万物)","publicationDate":null,"version":"Undated page, retrieved on 2026-10-15 (SHA-256 c6d11a3575e252ae47dc7548f4c1302dbd289897ab1aaf6c4d2f9b9f7829410d) directly from the publisher. Its wording can change; earlier wording is not authenticated.","canonicalUrl":"https://www.lingyiwanwu.com/about.html","evidenceUrl":"https://www.lingyiwanwu.com/about.html","language":"zh","archivedUrl":"https://web.archive.org/web/20261007110726/https://www.lingyiwanwu.com/about.html","archivedAt":"2026-10-07","accessedAt":"2026-10-15","reuse":{"attribution":"01.AI, “关于我们 — 01.AI 零一万物”.","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-gc-baidu-20f-2026","title":"Baidu, Inc. annual report on Form 20-F for the fiscal year ended December 31, 2025","authors":["Baidu, Inc."],"publisher":"U.S. Securities and Exchange Commission (EDGAR)","publicationDate":"2026-03-17","version":"Dated document, retrieved on 2026-10-15 (SHA-256 9b6bfb250d2268627cdfedfc586f3319b160d9a35567fd755d5bdaf52cbe6968) 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.sec.gov/Archives/edgar/data/1329099/000119312526109289/d38065d20f.htm","evidenceUrl":"https://www.sec.gov/Archives/edgar/data/1329099/000119312526109289/d38065d20f.htm","language":"en","accessedAt":"2026-10-15","reuse":{"attribution":"Baidu, Inc., “Baidu, Inc. annual report on Form 20-F for the fiscal year ended December 31, 2025”, 2026-03-17.","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-gs-rohin-shah-home","title":"Rohin Shah (personal homepage, undated)","authors":["Rohin Shah"],"publisher":"rohinshah.com","publicationDate":null,"version":"Undated post, retrieved on 2026-10-15 (SHA-256 77bc3df1eb53d569b61f13a1d38ab4902ceebca59e342ec6caaa1aa46ce669ab) directly from the publisher. Its wording can change; earlier wording is not authenticated.","canonicalUrl":"https://rohinshah.com/","evidenceUrl":"https://rohinshah.com/","language":"en","archivedUrl":"https://web.archive.org/web/20260917083329/https://rohinshah.com/","archivedAt":"2026-09-17","accessedAt":"2026-10-15","reuse":{"attribution":"Rohin Shah, “Rohin Shah (personal homepage, 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":"lg-src-gw-hcompany-about","title":"About us | H (company page, undated)","authors":["H Company"],"publisher":"H Company","publicationDate":null,"version":"Undated page, retrieved on 2026-10-15 (SHA-256 384a01b0115c61b29c5d688d1ca4a4e6a7b8a7da09c93a6ab157167f9567f481) directly from the publisher. Its wording can change; earlier wording is not authenticated.","canonicalUrl":"https://www.hcompany.ai/about","evidenceUrl":"https://www.hcompany.ai/about","language":"en","archivedUrl":"https://web.archive.org/web/20260929003747/https://hcompany.ai/about","archivedAt":"2026-09-29","accessedAt":"2026-10-15","reuse":{"attribution":"H Company, “About us | H (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":"lg-src-gw-sequoia-reflection-2025","title":"Partnering with Reflection: Toward Superintelligence, with Autonomous Coding","authors":["Stephanie Zhan","Charlie Curnin"],"publisher":"Sequoia Capital","publicationDate":"2025-03-07","version":"Dated page, retrieved on 2026-10-15 (SHA-256 5089a82bf34f432446896e540cfaee45f5bbbb9b8ff708724fe3f33578406dd0) directly from the publisher.","canonicalUrl":"https://sequoiacap.com/article/partnering-with-reflection-toward-superintelligence-with-autonomous-coding","evidenceUrl":"https://sequoiacap.com/article/partnering-with-reflection-toward-superintelligence-with-autonomous-coding","language":"en","archivedUrl":"https://web.archive.org/web/20261010210147/https://sequoiacap.com/article/partnering-with-reflection-toward-superintelligence-with-autonomous-coding","archivedAt":"2026-10-10","accessedAt":"2026-10-15","reuse":{"attribution":"Stephanie Zhan, Charlie Curnin, “Partnering with Reflection: Toward Superintelligence, with Autonomous Coding”, 2025-03-07.","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":"lg-src-gw-tc-h-2024","title":"French AI startup H raises $220M seed round","authors":["Romain Dillet"],"publisher":"TechCrunch","publicationDate":"2024-05-21","version":"Dated page, retrieved on 2026-10-15 (SHA-256 3c4d79349f2d9a33d6127ac538312938c246b256978de04fc574e474ccdc5102) directly from the publisher.","canonicalUrl":"https://techcrunch.com/2024/05/21/french-ai-startup-h-raises-220-million-seed-round/","evidenceUrl":"https://techcrunch.com/2024/05/21/french-ai-startup-h-raises-220-million-seed-round/","language":"en","archivedUrl":"https://web.archive.org/web/20260917114204/https://techcrunch.com/2024/05/21/french-ai-startup-h-raises-220-million-seed-round/","archivedAt":"2026-09-17","accessedAt":"2026-10-15","reuse":{"attribution":"Romain Dillet, “French AI startup H raises $220M seed round”, 2024-05-21.","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-gw-tc-weng-2026","title":"Thinking Machines co-founder Lilian Weng left the company citing health reasons, then joined OpenAI","authors":["Amanda Silberling"],"publisher":"TechCrunch","publicationDate":"2026-07-29","version":"Dated page, retrieved on 2026-10-15 (SHA-256 fc75c85543f0f742b2d8810c6b300051336090f681d3bf2cf0844066a6fef617) directly from the publisher.","canonicalUrl":"https://techcrunch.com/2026/07/29/thinking-machines-co-founder-lilian-weng-left-the-company-citing-health-reasons-then-joined-openai/","evidenceUrl":"https://techcrunch.com/2026/07/29/thinking-machines-co-founder-lilian-weng-left-the-company-citing-health-reasons-then-joined-openai/","language":"en","archivedUrl":"https://web.archive.org/web/20260922154459/https://techcrunch.com/2026/07/29/thinking-machines-co-founder-lilian-weng-left-the-company-citing-health-reasons-then-joined-openai/","archivedAt":"2026-09-22","accessedAt":"2026-10-15","reuse":{"attribution":"Amanda Silberling, “Thinking Machines co-founder Lilian Weng left the company citing health reasons, then joined OpenAI”, 2026-07-29.","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-oai-leadership-2022","title":"OpenAI leadership team update","authors":["Sam Altman"],"publisher":"OpenAI","publicationDate":"2022-05-05","version":"Dated page, retrieved on 2026-10-15 (SHA-256 ccdb6a29e5606701bc9836bb19884e131396ccec9a57f9a3c17f654ddfabd894) 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://openai.com/index/leadership-team-update/","evidenceUrl":"https://openai.com/index/leadership-team-update/","language":"en","archivedUrl":"https://web.archive.org/web/20260830045254/https://openai.com/index/leadership-team-update/","archivedAt":"2026-08-30","accessedAt":"2026-10-15","reuse":{"attribution":"Sam Altman, “OpenAI leadership team update”, 2022-05-05.","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-bletchley-declaration-2023","title":"The Bletchley Declaration by Countries Attending the AI Safety Summit, 1-2 November 2023","authors":["Countries attending the AI Safety Summit"],"publisher":"Department for Science, Innovation and Technology","publicationDate":"2023-11-01","version":"Retrieved on 2026-10-27 (SHA-256 6d62200d3fb1af286a04d0c7e2c9d97dfda5d19bf32dd577584eeb79295626a3). GOV.UK HTML policy paper as served on 2026-10-27, marked 'Updated 13 February 2025' (includes the New Zealand line). Publisher as printed on the inspected expression: GOV.UK (Prime Minister's Office, Foreign, Commonwealth & Development Office, Department for Science, Innovation and Technology).","canonicalUrl":"https://www.gov.uk/government/publications/ai-safety-summit-2023-the-bletchley-declaration/the-bletchley-declaration-by-countries-attending-the-ai-safety-summit-1-2-november-2023","evidenceUrl":"https://www.gov.uk/government/publications/ai-safety-summit-2023-the-bletchley-declaration/the-bletchley-declaration-by-countries-attending-the-ai-safety-summit-1-2-november-2023","language":"en","accessedAt":"2026-10-27","reuse":{"attribution":"Countries attending the AI Safety Summit, “The Bletchley Declaration by Countries Attending the AI Safety Summit, 1-2 November 2023”, 2023-11-01.","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-us-eo-14110-fr","title":"Executive Order 14110 of October 30, 2023: Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence (88 FR 75191)","authors":["The White House"],"publisher":"The White House","publicationDate":"2023-11-01","version":"Retrieved on 2026-10-27 (SHA-256 46739dfc1af256961458d3b73fa4d6f4b447b38eaa72efd80966463818be68ab). Federal Register print of 1 November 2023, Vol. 88, No. 210, pp. 75191–75226, FR Doc. 2023-24283. The PDF text layer renders superscript exponents inline (10^26 appears as \"1026\"); quotes stop before such numbers. Publisher as printed on the inspected expression: Federal Register (Office of the Federal Register, National Archives) via govinfo.gov.","canonicalUrl":"https://www.govinfo.gov/content/pkg/FR-2023-11-01/pdf/2023-24283.pdf","evidenceUrl":"https://www.govinfo.gov/content/pkg/FR-2023-11-01/pdf/2023-24283.pdf","language":"en","accessedAt":"2026-10-27","reuse":{"attribution":"The White House, “Executive Order 14110 of October 30, 2023: Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence (88 FR 75191)”, 2023-11-01.","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-ab-ostp-framework-2024","title":"Framework for Nucleic Acid Synthesis Screening (National Science and Technology Council, Fast Track Action Committee on Synthetic Nucleic Acid Procurement Screening)","authors":["Office of Science and Technology Policy, Executive Office of the President (copy served from the Biden White House archive)"],"publisher":"Office of Science and Technology Policy, Executive Office of the President (copy served from the Biden White House archive)","publicationDate":null,"version":"Retrieved for the candidate pack on 2026-10-28 (SHA-256 3b083af856f92dfa91bbf5de344ca81761a4eb4c3a0323fa390e37330a802715). The copy linked from the HHS ASPR page on 2026-10-28, marked on its cover 'Revised: September 2024' with '2024 Version 1: April 2024'. The April 2024 version 1 was not retrieved. Technical passages (screening windows, matching methods, the criteria that may inform which sequences are in scope) were read but are not cited. Retrieved again at admission on 2026-10-30 (SHA-256 3b083af856f92dfa91bbf5de344ca81761a4eb4c3a0323fa390e37330a802715, the same bytes).","canonicalUrl":"https://bidenwhitehouse.archives.gov/wp-content/uploads/2024/04/Nucleic-Acid_Synthesis_Screening_Framework.pdf","evidenceUrl":"https://bidenwhitehouse.archives.gov/wp-content/uploads/2024/04/Nucleic-Acid_Synthesis_Screening_Framework.pdf","language":"en","accessedAt":"2026-10-28","reuse":{"attribution":"Office of Science and Technology Policy, Executive Office of the President (copy served from the Biden White House archive), “Framework for Nucleic Acid Synthesis Screening (National Science and Technology Council, Fast Track Action Committee on Synthetic Nucleic Acid Procurement Screening)”.","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-brent-mckelvey-2025","title":"Contemporary AI foundation models increase biological weapons risk","authors":["RAND Corporation (working paper WR-A3853-1), posted to arXiv"],"publisher":"RAND Corporation (working paper WR-A3853-1), posted to arXiv","publicationDate":null,"version":"Retrieved for the candidate pack on 2026-10-28 (SHA-256 1cc64de0a880543bda08d64374ee47539c029c55b187ab3f4e66603e3f8ad7e5). arXiv v1 of 12 June 2025; the PDF is the RAND working paper WR-A3853-1 of June 2025, marked \"not been formally reviewed or edited\" Retrieved again at admission on 2026-10-30 (SHA-256 1cc64de0a880543bda08d64374ee47539c029c55b187ab3f4e66603e3f8ad7e5, the same bytes).","canonicalUrl":"https://arxiv.org/pdf/2506.13798v1","evidenceUrl":"https://arxiv.org/pdf/2506.13798v1","language":"en","accessedAt":"2026-10-28","reuse":{"attribution":"RAND Corporation (working paper WR-A3853-1), posted to arXiv, “Contemporary AI foundation models increase biological weapons risk”.","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-arxiv-2506-13798-abs","title":"arXiv abstract page for 2506.13798 (submission history and report number)","authors":["arXiv (Cornell University)"],"publisher":"arXiv (Cornell University)","publicationDate":null,"version":"Retrieved for the candidate pack on 2026-10-28 (SHA-256 64847d8461f1f0361daccfb569e8d6cc053bfd8a4125c5339eeda6ec8bb06c17). Abstract page as served on 2026-10-28; used only for the v1 submission date and the RAND report number Retrieved again at admission on 2026-10-30 (SHA-256 63834668f90e3a0e80b98abc1ee48e42834a3ecc9060acc685a1a048bafb24e2). Its wording can change; earlier wording is not authenticated.","canonicalUrl":"https://arxiv.org/abs/2506.13798","evidenceUrl":"https://arxiv.org/abs/2506.13798","language":"en","accessedAt":"2026-10-28","reuse":{"attribution":"arXiv (Cornell University), “arXiv abstract page for 2506.13798 (submission history and report number)”.","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-vct-abs","title":"arXiv abstract page for arXiv:2504.16137 (Virology Capabilities Test (VCT): A Multimodal Virology Q&A Benchmark)","authors":["arXiv (Cornell University)"],"publisher":"arXiv (Cornell University)","publicationDate":null,"version":"Retrieved for the candidate pack on 2026-10-28 (SHA-256 3ad6043cc84942ab180140c67ad4e128b37899ee1e4fc957d8f80d93a80d4898). Abstract page as served on 2026-10-28, listing v1 of 21 April 2025 and v2 of 29 April 2025 (the current version). Only the abstract page was read; the full paper was not retrieved, so no record rests on its body. Retrieved again at admission on 2026-10-30 (SHA-256 cc6f6960f2503c7789a77800e49cac2de3af708305ca7f232a5322507a12bc80). Its wording can change; earlier wording is not authenticated.","canonicalUrl":"https://arxiv.org/abs/2504.16137","evidenceUrl":"https://arxiv.org/abs/2504.16137","language":"en","accessedAt":"2026-10-28","reuse":{"attribution":"arXiv (Cornell University), “arXiv abstract page for arXiv:2504.16137 (Virology Capabilities Test (VCT): A Multimodal Virology Q&A Benchmark)”.","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-good-authorship-1965","sourceRole":"primary","support":"supports","locator":"Scanned reprint, page 1, author byline","passage":"The paper names I. J. Good as its author.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-09-17","note":"Source passage inspected by an automated research assistant. Human editorial review is pending."},{"id":"evidence-good-catalogue","sourceRole":"secondary","support":"supports","locator":"dc.contributor.author and dc.description.notes; distinguish dc.date.issued","passage":"The repository attributes the paper to Good and identifies a reprint from volume 6 with a 1965 copyright. Its issued date of 2005 applies to the repository technical report.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-09-17","note":"Catalogue metadata corroborates identity and date; it does not independently establish the technical propositions. Human review pending."},{"id":"evidence-chris-olah-cofounded-anthropic-2021","sourceRole":"primary","support":"supports","locator":"PDF page 6, “Our business”, Founded panel and Chris Olah row","passage":"The 2021 founding panel credits Chris Olah as a co-founder.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-09-18","note":"Relevant source passage inspected by an automated research assistant. Human source review is pending."},{"id":"evidence-atlas-gdm-gemini-report","sourceRole":"primary","support":"supports","locator":"Introducing Gemini section, by Demis Hassabis; opening paragraphs and Gemini technical report link","passage":"Hassabis introduces Gemini as Google DeepMind CEO on behalf of the Gemini team and links the technical report.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-09-29","note":"Rechecked publisher identity against the original launch article. The arXiv author is the cross-Google Gemini team; this claim concerns the dated institutional introduction, not sole authorship of the report. No human review recorded."},{"id":"evidence-atlas-feedback-cites-paper","sourceRole":"primary","support":"supports","locator":"Paragraph beginning “The system”, linked paper title","passage":"The account explicitly links the human-preferences paper.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-09-28","note":"Original source passage checked by the machine agent that proposed the record. No independent machine audit or human source inspection is recorded for this increment."},{"id":"evidence-prov-ms-fgf-2025-low-medium-deployment","sourceRole":"primary","support":"supports","locator":"PDF p. 7 (printed p. 6), §3 Evaluation and assessment, Deeper capability assessment, paragraph 'Pre-mitigation capability assessment', second sentence","passage":"Models assessed as posing low or medium risk may be deployed with appropriate safeguards as outlined below.","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 2025-02-26. Extraction note: Appendix I gives the deployment requirement 'Deployment allowed in line with Responsible AI Program requirements' for the Low and Medium rows. Human editorial review is pending."},{"id":"evidence-ts-b5-wiener-1960-context","sourceRole":"primary","support":"supports","locator":"Printed p. 1357, paragraphs before the heading \"Man and Slave\"","passage":"if the rules for victory in a war game do not correspond to what we actually wish for our country, it is more than likely that such a machine may produce a policy which would win a nominal victory on points","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-06","note":"The same passage as the B6 record."},{"id":"evidence-ts-a3-clemmensen-1996-08-30-context","sourceRole":"primary","support":"supports","locator":"Message body, first unquoted paragraph","passage":"my belief is that there is really only one relevant SI, and that is an SI whose motivation is to become more intelligent. This SI is the important one, because this is the one that has a built-in positive feedback mechanism.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-06","note":"The same passage as the A1 record."},{"id":"evidence-ts-a4-clemmensen-1996-09-07","sourceRole":"primary","support":"supports","locator":"Message body, second and final unquoted paragraphs (\"obvous\" sic)","passage":"the SI cannot instantly stop all wars without revealing itself [...] an SI that's only somewhat more intelligent than a smart human can make itself smarter still [...] the transformation will occur in days or weeks.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-06","note":"Checked against the archive HTML on 2026-10-06 (hash matched the 1996–2022 candidate search)."},{"id":"evidence-ts-a4-sits-context","sourceRole":"primary","support":"supports","locator":"\"The short version\", opening of the page","passage":"If computing speeds double every two years, what happens when computer-based AIs are doing the research? [...] Two years after Artificial Intelligences reach human equivalence, their speed doubles. One year later, their speed doubles again. Six months - three months - 1.5 months ... Singularity.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-06","note":"The same passage as the A5 record."},{"id":"evidence-ts-a6-christiano-2018-takeoff-contests","sourceRole":"primary","support":"supports","locator":"Section \"Slow takeoff\", paragraph before \"Operationalizing slow takeoff\"","passage":"Being 12 months ahead in AGI might get you a decisive strategic advantage, but the world has accelerated so much that that’s just about as hard as getting to airplanes 30 years before anyone else.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-06","note":"Checked against the current page on 2026-10-06; the 1996–2022 candidate search cites this sentence in its note."},{"id":"evidence-yk-t03-s1","sourceRole":"primary","support":"supports","locator":"message body, paragraph beginning 'But that's only one third of the problem'","passage":"But that's only one third of the problem. First, ve has to *want* to know whether that sinking sensation will result. Second, ve has to model it accurately. Third, ve has to change vis behavior based on that model.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-07","note":"Candidate pack passage T03-s1. The quotation (39 words) matched the retrieved expression (SHA-256 4fb349f96304ea9c661f22830b6831e3a8dba94ea90584ebabcf9a4ff7e46263) when checked on 2026-10-07."},{"id":"evidence-yk-t19-s2","sourceRole":"primary","support":"supports","locator":"message body, paragraph beginning 'Evil nanotech?'","passage":"Evil is a term that applies to cognitive systems, not systems of a particular size. And an UFAI is not evil; it just transforms local reality according to a decision criterion that makes no internal mention of human life.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-07","note":"Candidate pack passage T19-s2. The quotation (39 words) matched the retrieved expression (SHA-256 7c6b98d1ab322db0e7bdbcc1cb9297921bb54f76781dae7865eaf188711aab56) when checked on 2026-10-07."},{"id":"evidence-yk-t26-e1","sourceRole":"primary","support":"supports","locator":"message body, paragraph beginning 'Part one: If it's even…'","passage":"will know *exactly* what to say","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-07","note":"Candidate pack passage T26-e1. The quotation (6 words; trimmed 2026-10-13 from 27 words) matched the retrieved expression (SHA-256 ec8f3d5b7d58bd04717ee7fe30f687d324c7d5168b51207549e834a3ef0062b0) when checked on 2026-10-07. The dataset already quotes this source for the human-thought collection; together the quotations would exceed the per-document quotation limit. Shortened to a verbatim fragment of the pack's quotation; the condition it omits is in the record's statement."},{"id":"evidence-ec-m06a-03","sourceRole":"primary","support":"supports","locator":"printed p. 620 (statement, 'From technology to economics')","passage":"It is evident that it will change wage income both by number of jobs — some places up and some places down","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-14","note":"Candidate pack passage M06a-03 (research/economics-mainstream). The quotation (22 words) matched the retrieved current copy (SHA-256 3962809fbb923be62bf986addb0014ee147934a8767475c1f7cfe051959bff90) in the pack's verification. Re-checked at admission on 2026-10-14 against a copy retrieved that day."},{"id":"lg-ev-oai-brockman-president-2022","sourceRole":"primary","support":"supports","locator":"Paragraph beginning “Greg Brockman”","passage":"Greg Brockman is becoming President, a new role","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The quotation (8 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-cn-01ai-founded-2023","sourceRole":"primary","support":"supports","locator":"关于我们, 公司介绍 (Company introduction), first sentence","passage":"零一万物成立于2023年5月，总部位于北京，由李开复博士领衔创立","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The quotation (1 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. No human editorial review.","retrievalRoute":"direct"},{"id":"lg-ev-mx-wu-xai-team-2023","sourceRole":"secondary","support":"supports","locator":"Paragraph beginning 'That list includes'","passage":"Ars Technica lists Tony Wu, of the University of Toronto, among the team named on xAI's website.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The paraphrase’s verbatim anchor “Tony Wu (University of Toronto)” 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. 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":"lg-ev-ant-krieger-cpo-2024","sourceRole":"primary","support":"supports","locator":"First paragraph","passage":"Mike Krieger has joined Anthropic as our Chief Product Officer","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The quotation (10 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":"lg-ev-ant-krieger-cpo-2024-labs","sourceRole":"primary","support":"context","locator":"Paragraph beginning “Today we’re building on this approach” (13 January 2026)","passage":"has spent the past two years as Anthropic’s Chief Product Officer","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The quotation (11 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":"lg-ev-gw-weng-openai-return-2026-a","sourceRole":"secondary","support":"supports","locator":"Paragraph 3","passage":"OpenAI told TechCrunch that Weng would be rejoining the company","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The quotation (10 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 judged it partly supported; its findings were applied at admission. No human editorial review.","retrievalRoute":"direct"},{"id":"lg-ev-gw-weng-openai-return-2026-b","sourceRole":"secondary","support":"supports","locator":"Paragraph beginning “According to a company spokesperson”","passage":"Weng will lead a top-level team focused on accelerating OpenAI’s internal research","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 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 judged it partly supported; its findings were applied at admission. No human editorial review.","retrievalRoute":"direct"},{"id":"lg-ev-gs-shah-gdm-asat-lead","sourceRole":"primary","support":"supports","locator":"Home, paragraph 1","passage":"I lead the AGI Safety & Alignment team at Google DeepMind","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The quotation (11 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. No human editorial review.","retrievalRoute":"direct"},{"id":"lg-ev-gw-antonoglou-cofounded-reflection","sourceRole":"secondary","support":"supports","locator":"Photo caption below the byline","passage":"Reflection co-founders Misha Laskin and Ioannis Antonoglou.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The quotation (7 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":"lg-ev-gw-kantor-cofounded-h-company","sourceRole":"secondary","support":"supports","locator":"Paragraph beginning “It has managed to raise”","passage":"Charles Kantor, the startup’s co-founder and CEO, was a university researcher at Stanford.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The quotation (13 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 judged it partly supported; its findings were applied at admission. No human editorial review.","retrievalRoute":"direct"},{"id":"lg-ev-gw-kantor-cofounded-h-company-cloix","sourceRole":"primary","support":"context","locator":"Milestones list, entry for 2025","passage":"Gautier Cloix becomes CEO.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The quotation (4 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 judged it partly supported; its findings were applied at admission. No human editorial review.","retrievalRoute":"direct"},{"id":"ev-cmp-rsp-v2-1-d6-competitor-1","sourceRole":"primary","support":"supports","locator":"PDF p. 17 (printed p. 13), footnote 17 (attached to §7.1 item 7 'Policy changes'), first and second sentences","passage":"It is possible at some point in the future that another actor in the frontier AI ecosystem will pass, or be on track to imminently pass, a Capability Threshold without implementing measures equivalent to the Required Safeguards","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-27","note":"Verbatim excerpt checked by an automated research assistant on 2026-10-27 against the publisher copy, whose SHA-256 matches the copy retrieved on 2026-10-15. The footnote continues that in such a scenario Anthropic might decide to lower its Required Safeguards, would acknowledge the overall risk, and would make a case to the U.S. government for regulatory action. Human editorial review is pending."},{"id":"ev-cmp-rsp-v2-1-d6-discretion-1","sourceRole":"primary","support":"supports","locator":"PDF p. 16 (printed p. 12), §7.1, footnote 16","passage":"where this policy is unintentionally ambiguous, we will act in accordance with the Responsible Scaling Officer or CEO’s judgment","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-27","note":"Verbatim excerpt checked by an automated research assistant on 2026-10-27 against the publisher copy, whose SHA-256 matches the copy retrieved on 2026-10-15. Other stated discretion: CEO and RSO may approve interim measures (§6.2, PDF p. 15); trusted users may receive a version with reduced safeguards under alternative controls (§4.1 item 6, PDF p. 12); the RSO may limit further training below the default 4x (footnote 8, PDF p. 11). Human editorial review is pending."},{"id":"evidence-ap-ajeya-cotra-cotra-2022-without-specific-countermeasures","sourceRole":"primary","support":"supports","locator":"Post author list in the publisher’s post data (LessWrong GraphQL API)","passage":"Ajeya Cotra","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 through the publisher’s post data and checked by a reviewer from a different model family on 2026-10-27. Human editorial review is pending.","retrievalRoute":"direct"},{"id":"evidence-ap-vladimir-mikulik-rlo-2019-introduction","sourceRole":"primary","support":"supports","locator":"Post author list in the publisher’s post data (LessWrong GraphQL API)","passage":"Vlad Mikulik","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 through the publisher’s post data and checked by a reviewer from a different model family on 2026-10-27. Human editorial review is pending.","retrievalRoute":"direct"},{"id":"evidence-ap-eliezer-yudkowsky-lw-2008-ghosts-in-the-machine","sourceRole":"primary","support":"supports","locator":"Post author list in the publisher’s post data (LessWrong GraphQL API)","passage":"Eliezer 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 through the publisher’s post data. Human editorial review is pending.","retrievalRoute":"direct"},{"id":"ev-cmp-rsp-v2-1-d1-cbrn-1","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 edition’s CBRN-3 Capability Threshold is the ability to significantly help people with basic technical backgrounds create or obtain and deploy CBRN weapons.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-27","note":"Explicit paraphrase, written on 2026-10-27 by an automated research assistant after re-reading the passage in the publisher copy, whose SHA-256 matches the copy retrieved on 2026-10-15. The comparison quoted the passage; it is given here as a paraphrase to keep the edition within the quotation limit. Undergraduate STEM degrees are given as an example of a basic technical background. A second threshold, CBRN-4, concerns substantial uplift to moderately resourced state programs (same table, PDF p. 8; Appendix C, PDF p. 20). Human editorial review is pending."},{"id":"ev-cmp-rsp-v2-1-d7-external-1","sourceRole":"primary","support":"supports","locator":"PDF p. 17 (printed p. 13), §7.2 Transparency and External Input, numbered item 1 ('Public disclosures'), first sentence","passage":"The edition says Anthropic will publicly release key information on the evaluation and deployment of its models, excluding sensitive details.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-27","note":"Explicit paraphrase, written on 2026-10-27 by an automated research assistant after re-reading the passage in the publisher copy, whose SHA-256 matches the copy retrieved on 2026-10-15. The comparison quoted the passage; it is given here as a paraphrase to keep the edition within the quotation limit. §7.2 item 3 commits to inform a relevant U.S. Government entity when a model needs protections beyond the ASL-2 Standard; footnote 18 expects evaluation details to go to a U.S. Government entity where a model is not deployed publicly. Human editorial review is pending."},{"id":"ev-cmp-fsf-v2-0-d2-levels-1","sourceRole":"primary","support":"supports","locator":"PDF p. 2 (printed p. 2), Framework, component '1 - Critical Capability Levels', second paragraph","passage":"The edition sets out two groups of CCLs: misuse CCLs, which can signal a heightened risk that misuse causes severe harm if left unaddressed, and deceptive alignment CCLs.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-27","note":"Explicit paraphrase, written on 2026-10-27 by an automated research assistant after re-reading the passage in the publisher copy, whose SHA-256 matches the copy retrieved on 2026-10-15. The comparison quoted the passage; it is given here as a paraphrase to keep the edition within the quotation limit. Human editorial review is pending."},{"id":"ev-cmp-frontier-governance-2026-d8-process-1","sourceRole":"primary","support":"supports","locator":"PDF p. 21 (printed p. 19), §7.1 Update and approval process, fourth (final) sentence, up to 'for oversight'","passage":"The edition says material updates will be presented for oversight to the OpenAI Foundation board's Safety and Security Committee and to OpenAI Ireland Limited's board of directors.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-27","note":"Explicit paraphrase, written on 2026-10-27 by an automated research assistant after re-reading the passage in the publisher copy, whose SHA-256 matches the copy retrieved on 2026-10-15. The comparison quoted the passage; it is given here as a paraphrase to keep the edition within the quotation limit. Human editorial review is pending."},{"id":"ev-cmp-frontier-governance-2026-d8-process-2","sourceRole":"primary","support":"supports","locator":"PDF p. 22 (printed p. 20), §7.2 Framework assessment, first sentence","passage":"The edition says a Framework Assessment will be completed by OpenAI at least every 12 months, counted from when the EU Code of Practice and the TFAIA took effect.","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 that the 12 months run from the effective dates of the TFAIA and the EU Code of Practice. 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 same sentence also bases the assessment on factors that OpenAI’s legal function considers, for example legal changes or changes in model capabilities. Human editorial review is pending."},{"id":"ev-cmp-faif-2025-d5-cadence-1","sourceRole":"primary","support":"supports","locator":"PDF p. 10, 'Operational and Societal Risks' › '5. Deployment Decisions', second paragraph, third and fourth sentences","passage":"The edition says that reviews before deployment cover benchmark results, such as WMDP scores, and how effective mitigations are, and that for internal use xAI reviews catastrophic risks such as oversight evasion before a broad rollout.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-27","note":"Explicit paraphrase, written on 2026-10-27 by an automated research assistant after re-reading the passage in the publisher copy, whose SHA-256 matches the copy retrieved on 2026-10-15. The comparison quoted the passage; it is given here as a paraphrase to keep the edition within the quotation limit. Human editorial review is pending."},{"id":"c2-ev-baidu-20f-fy2025-ernie-5","sourceRole":"primary","support":"supports","locator":"Item 4.B, Business overview, 'AI Capabilities'","passage":"In November 2025, we unveiled ERNIE 5.0, our first native omni-modal foundation model at Baidu World 2025.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-27","note":"Candidate pack passage, Chinese frontier-AI institutions, second collection (research/china-labs-2). The quotation (17 words) matched the retrieved text on 2026-10-27 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 (direct retrieval failed); 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":"c2-ev-baidu-20f-fy2025-ernie-5-update","sourceRole":"primary","support":"supports","locator":"Item 4.B, Business overview, 'AI Capabilities'","passage":"In January 2026, we released an updated version of ERNIE 5.0.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-27","note":"Candidate pack passage, Chinese frontier-AI institutions, second collection (research/china-labs-2). The quotation (11 words) matched the retrieved text on 2026-10-27 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 (direct retrieval failed); 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":"evidence-ap-katja-grace-grace-2014-superintelligence-reading-group","sourceRole":"primary","support":"supports","locator":"“by” line beneath the title (username), linked to the account page /users/katjagrace","passage":"KatjaGrace","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 the current post page, whose author link was followed (research/author-persons/codex-identity-2026-10-27.md); the preparer had restated the source record’s author list in place of the username. Human editorial review is pending.","retrievalRoute":"direct"},{"id":"evidence-fl-link-rsp-v3-1-sb53-1","sourceRole":"primary","support":"supports","locator":"Footnote 1 (Introduction), PDF p. 4","passage":"Where laws such as California SB 53 define this or similar terms with specific thresholds, we address those requirements in separate compliance frameworks.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-28","note":"Candidate pack passage, frameworks to law (research/frameworks-to-law). The quotation (23 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-cdn.anthropic.com. Matching ignored differences of whitespace, case, quotation marks, dashes and Unicode compatibility forms, so the source’s typography can differ. Same sentence as in v3.0. A cross-family machine check (2026-10-28, item 115) re-read the record’s sources and found full support. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-fl-intl-bletchley-developer-responsibility-1","sourceRole":"primary","support":"supports","locator":"Paragraph beginning 'We affirm that, whilst safety must be considered across the AI lifecycle'","passage":"actors developing frontier AI capabilities, in particular those AI systems which are unusually powerful and potentially harmful, have a particularly strong responsibility for ensuring the safety of these AI systems, including through systems for safety testing, through evaluations, and by other appropriate measures","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-28","note":"Candidate pack passage, frameworks to law (research/frameworks-to-law). The quotation (43 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 of the same address (the page’s bytes had changed). Read directly from www.gov.uk. 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-ab-scr-ostp-framework-attestation-1","sourceRole":"primary","support":"supports","locator":"V. Screening Framework, criterion 1, p. 8","passage":"either post on a public webpage a statement that self-attests to following each of the criteria in this framework, or provide such a statement to the federally funded customer and the federal funding agency upon request","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-30","note":"Candidate pack passage, AI and biosecurity (research/ai-bio). The quotation (36 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. A cross-family machine check (2026-10-30) re-read this record outside its brief, with the other records resting on the same source, and agreed in part; its corrections were applied at admission. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-ab-scr-ostp-framework-attestation-2","sourceRole":"primary","support":"supports","locator":"V. Screening Framework, criterion 1, items (i) to (iv), p. 8","passage":"the attestation should be signed by an individual with authority to respond for the organisation, give point-of-contact information, be updated by 1 January each year or sooner if contacts change, and expressly commit the organisation, if it ceases to adhere, to notify within 72 hours any federally funded customers and funding agencies to which it had attested.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-30","note":"Candidate pack passage, AI and biosecurity (research/ai-bio). 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. Explicit paraphrase of the four lettered items. The framework adds that OSTP anticipated coordinating a standardised template statement. A cross-family machine check (2026-10-30) re-read this record outside its brief, with the other records resting on the same source, and agreed in part; its corrections were applied at admission. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-ab-gov-us-eo14110-sec41b-doe-tools-1","sourceRole":"primary","support":"supports","locator":"Sec. 4.1(b), 88 FR 75196","passage":"the Secretary shall develop tools to evaluate AI capabilities to generate outputs that may represent nuclear, nonproliferation, biological, chemical, critical infrastructure, and energy-security threats or hazards","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-30","note":"Candidate pack passage, AI and biosecurity (research/ai-bio). The quotation (26 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. A cross-family machine check (2026-10-30, item 81) re-read the record’s sources and found full support. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-ab-gov-us-eo14110-sec41b-doe-tools-2","sourceRole":"primary","support":"supports","locator":"Sec. 4.1(b) (continued), 88 FR 75197","passage":"do this work solely for the purposes of guarding against these threats, and shall also develop model guardrails that reduce such risks","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-30","note":"Candidate pack passage, AI and biosecurity (research/ai-bio). The quotation (22 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. A cross-family machine check (2026-10-30, item 81) re-read the record’s sources and found full support. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-ab-ev-vct-benchmark-measures-1","sourceRole":"primary","support":"supports","locator":"Abstract, first two sentences","passage":"a large language model (LLM) benchmark that measures the capability to troubleshoot complex virology laboratory protocols","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-30","note":"Candidate pack passage, AI and biosecurity (research/ai-bio). The quotation (16 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 cc6f6960f2503c7789a77800e49cac2de3af708305ca7f232a5322507a12bc80; 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. The abstract adds that the benchmark was constructed from the inputs of dozens of PhD-level expert virologists. A cross-family machine check (2026-10-30, item 27) re-read the record’s sources and found full support. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-ab-arg-d2-brent-contests-1","sourceRole":"primary","support":"supports","locator":"Summary, p. iv (PDF p. 4)","passage":"First, we argue they mistakenly assume biological weapons development requires tacit knowledge – expertise gained through experience that cannot be conveyed in words, including by large language models (LLMs).","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-30","note":"Candidate pack passage, AI and biosecurity (research/ai-bio). The quotation (28 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 paper is the primary source for its own argument. A cross-family machine check (2026-10-30, item 116) re-read the record’s sources and found full support. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-ab-arg-d2-brent-contests-2","sourceRole":"primary","support":"supports","locator":"Submission history block, v1 entry","passage":"[v1] Thu, 12 Jun 2025","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-30","note":"Candidate pack passage, AI and biosecurity (research/ai-bio). The quotation (5 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 63834668f90e3a0e80b98abc1ee48e42834a3ecc9060acc685a1a048bafb24e2; 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 day of the date rested on no cited passage. Cited for the submission date only; the retrieved PDF carries no arXiv stamp and its cover gives only the month. A cross-family machine check (2026-10-30, item 116) re-read the record’s sources and found full support. No human editorial review.","retrievalRoute":"direct"}],"at":{"entities":[0,1,2,3,19,20,24,30,43,46,50,55,56,67,92,144,145,201,205,249,262,286,293,294,295,320,327,330,335,338,339,341,345,352,368,374,430,453,482,518,567,635,645,679,690,729,744,773,780,835,867,900,939,1006,1018,1038,1219,1235,1252,1255,1266],"claims":[23,37,312,316,476,782,860,876,881,952,1009,1118,1159,1391,1624,1647,1660,1687,1892,1918,1977,1989,2058,2059,2213,2290,2438,2505,2514,2605,2669,2677,2832,2861,2924,2935,3305,3329,3380,3432],"sources":[11,14,17,22,37,40,45,50,51,69,140,183,262,287,294,295,297,325,332,335,340,397,420,449,520,567,568,577,610,663,675,684,687,690,748,848,874,1047,1076,1077,1091],"evidence":[24,29,39,334,338,507,992,1074,1090,1096,1175,1236,1358,1407,1737,2047,2075,2090,2130,2131,2442,2443,2481,2554,2568,2569,2659,2660,2860,2937,3085,3152,3162,3271,3348,3349,3358,3615,3616,3664,3778,3794,4534,4535,4574,4575,4643,4734,4735]},"version":"1lg47t7"}