{"entities":[{"id":"openai","name":"OpenAI"},{"id":"anthropic","name":"Anthropic"},{"id":"arc-evals","name":"ARC Evals"},{"id":"eliezer-yudkowsky","name":"Eliezer Yudkowsky"},{"id":"rsp-v2-0","name":"Anthropic Responsible Scaling Policy — v2.0"},{"id":"rsp-v3-0","name":"Anthropic Responsible Scaling Policy — v3.0"},{"id":"rsp-v3-3","name":"Anthropic Responsible Scaling Policy — v3.3"},{"id":"rsp-v3-4","name":"Anthropic Responsible Scaling Policy — v3.4"},{"id":"preparedness-beta-2023","name":"OpenAI Preparedness Framework — Beta"},{"id":"arc-pilot-announcement-2023","name":"ARC Evals autonomous-task pilot announcement (2023)"},{"id":"gpt4-system-card-2023","name":"GPT-4 System Card"},{"id":"meta-faf-v1-1","name":"Meta Frontier AI Framework — Version 1.1"},{"id":"openai-inc","name":"OpenAI, Inc. (OpenAI Foundation)"},{"id":"inflection-ai","name":"Inflection AI"},{"id":"extropians-1996-12-31-yudkowsky-5142","name":"Re: Goal-based AI"},{"id":"rlo-2019-deceptive-alignment","name":"Deceptive Alignment (Risks from Learned Optimization, part 4)"},{"id":"treacherous-turn","name":"Treacherous turn"},{"id":"intelligence-brings-benevolence","name":"Sufficient intelligence brings good values"},{"id":"short-instruction-suffices","name":"A short instruction such as making everyone happy"},{"id":"warning-signs-in-time","name":"Clear warning signs will come before danger"},{"id":"extropians-1998-12-06-yudkowsky-2153","name":"Singularity: AI Morality"},{"id":"lw-2008-fighting-a-rearguard-action-against-the","name":"Fighting a Rearguard Action Against the Truth"},{"id":"lw-2008-that-tiny-note-of-discord","name":"That Tiny Note of Discord"},{"id":"sl4-2005-07-13-yudkowsky-0507-11477","name":"Re: Fighting UFAI"},{"id":"eu-ai-act-2024","name":"Regulation (EU) 2024/1689 (Artificial Intelligence Act), as published in the Official Journal"},{"id":"explosive-growth-estimates","name":"Estimates of explosive growth within a stated horizon"},{"id":"ai-as-task-automation","name":"AI as the automation of tasks"},{"id":"recent-economics-engagement-with-ai","name":"Recent entry of machine intelligence into mainstream economics"},{"id":"acemoglu-2024-simple-macroeconomics-of-ai","name":"The Simple Macroeconomics of AI"},{"id":"korinek-stiglitz-2017-ai-income-distribution","name":"Artificial Intelligence and Its Implications for Income Distribution and Unemployment"},{"id":"acemoglu-restrepo-2018-modeling-automation","name":"Modeling Automation"},{"id":"ben-mann","name":"Ben Mann"},{"id":"hannah-pritchett","name":"Hannah Pritchett"},{"id":"helen-toner","name":"Helen Toner"},{"id":"vibhu-mittal","name":"Vibhu Mittal"},{"id":"evan-hubinger","name":"Evan Hubinger"},{"id":"amd","name":"Advanced Micro Devices, Inc. (AMD)"},{"id":"openai-gpt5-system-card-2025","name":"GPT-5 System Card"}],"claims":[{"id":"rsp-v3-4-coverage-date","statement":"RSP v3.4 allows a Risk Report coverage date up to 30 days before publication.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Report coverage and publication dates are distinct. This claim does not say every report uses the maximum lag.","evidenceIds":["evidence-rsp-v3-4-coverage-date"],"attributes":{"changeBasis":"changelog"}},{"id":"eval-i-openai-gpt4-card","statement":"OpenAI published the GPT-4 system card.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Version uncertainty is retained; the printed pagination starts at 41.","evidenceIds":["evidence-eval-i-openai-gpt4-card"]},{"id":"eval-i-arc-pilot","statement":"ARC Evals published an account of its autonomous-task pilot.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Displayed publication date differs from the URL date. Tests of scaffolded agents are not bounds on every possible use of the underlying models.","evidenceIds":["evidence-eval-i-arc-pilot"]},{"id":"prov-preparedness-beta-2023-safety-enhancing-development-exception","statement":"Preparedness Framework (Beta) states that its development restriction should not preclude safety-enhancing development.","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-preparedness-beta-2023-safety-enhancing-development-exception"],"attributes":{"provision":{"kind":"exception","key":"pf-safety-enhancing-development-exception"}}},{"id":"ts-b1-yudkowsky-1996-12-31-5142-context","statement":"Yudkowsky’s message of 1996-12-31 says a confined evil Power could lie to be let out.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The message concerns release from confinement.","evidenceIds":["evidence-ts-b1-yudkowsky-1996-12-31-5142-context"],"attributes":{"thought":{"roles":["context"],"register":"nonfiction","missing":"Persuasive lying to gain release is stated, but not a later change of behaviour once the Power is free or strong."}}},{"id":"yk-y02-lw-2008-that-tiny-note-of-discord-c","statement":"The post “That Tiny Note of Discord” (2008-09-23) contests Y02. Recapping his 2008 autobiographical series, he states that his 1997 self believed any superintelligence would automatically do what is right, and names the flaw in that belief.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Premise addressed: Greater intelligence by itself brings benevolent goals. Imported Overcoming Bias post (2008-09-23). In the HTML the name is written Eliezer<sub>1997</sub>; the quote gives the tag-stripped text. Clearest single retrospective statement of the earlier belief found; companion posts are listed in searchNotes.","evidenceIds":["evidence-yk-t02-e"],"attributes":{"thought":{"roles":["contesting","later-restatement"],"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 greater intelligence by itself brings benevolent goals."},"retrospective":{"from":"1997","until":"1997","note":"Reports the belief of “Eliezer1997”."}}}},{"id":"yk-y04-extropians-1998-12-06-yudkowsky-2153-c","statement":"The Extropians message “Singularity: AI Morality” (1998-12-06) contests Y04. Replying to Nick Bostrom, who proposed building a moral system with particular values into an AI, he argues that a simple built-in value such as human survival, with a rigid definition, can have unforeseen and horrific consequences (the passage goes on to describe endless confinement).","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Premise addressed: A short instruction such as 'make people happy' specifies what we want. The instruction discussed is 'a large value on human survival', not happiness. A hypothetical illustration. His 1998 alternative was to program the AI to 'seek out *correct* answers' (objective morality), a position he later abandoned (see ‘Sufficient intelligence brings good values’, Y02).","evidenceIds":["evidence-yk-t04-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 a short instruction such as 'make people happy' specifies what we want."}}}},{"id":"yk-y08-sl4-2005-07-13-yudkowsky-0507-11477-c","statement":"The SL4 message “Re: Fighting UFAI” (2005-07-13) contests Y08. Replying to Justin Corwin in a thread on fighting an unfriendly AI, he calls speculation about drastic countermeasures largely pointless because no government body would understand the threat or have the authority to act on it.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Premise addressed: Once a warning sign is seen, it will lead to effective action. Concerns the capacity to respond to an unfriendly AI, not warning signs as such; fit to the premise is partial.","evidenceIds":["evidence-yk-t08-s6"],"attributes":{"thought":{"roles":["contesting"],"register":"nonfiction","engagement":{"kind":"contrary-only","rationale":"Under the message-level rule, the passage states the contrary conclusion without giving a reason against the premise that once a warning sign is seen, it will lead to effective action, even when read with its message or dialogue turn."}}}},{"id":"aia-art111-earlier-models","statement":"Article 111(3) provides that providers of general-purpose AI models that have been placed on the market before 2 August 2025 shall take the necessary steps to comply with the obligations laid down in the Regulation by 2 August 2027.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"A transitional date for models already on the market. The text does not identify any particular model; whether any named model was placed on the market before the date is a matter of fact and legal application that is not recorded. Regulation (EU) 2026/1744 amends other paragraphs of Article 111 (2 and a new 4); in the inspected text, paragraph 3 is not amended.","evidenceIds":["evidence-aia-art111-earlier-models-1","evidence-aia-art111-earlier-models-2"],"attributes":{"provision":{"kind":"required_action","key":"aia-gpai-earlier-models-transition"}}},{"id":"ec-g09-acemoglu-2024-simple-macroeconomics-of-ai-b","statement":"Acemoglu, “The Simple Macroeconomics of AI” (2024-04-05) bears on G09. Compares his ten-year estimate with larger forecasts. A footnote cites Davidson (2021) among those who expect \"explosive\" growth.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The record documents what the text says; it does not establish that the argument or model is right.","evidenceIds":["evidence-ec-g09-i"],"attributes":{"thought":{"roles":["context"],"register":"nonfiction","missing":"Concerns the next ten years and current AI; does not assess explosive growth this century.","attested":{"date":"2025-01-06","note":"Only retrieved capture of the PDF."},"tier":{"value":1,"basis":"Tier 1 (unchanged). Economist (MIT Economics, author-hosted PDF) writing for a core economics venue: prepared for Economic Policy; NBER Working Paper 32487; journal version in Economic Policy."}}}},{"id":"ec-m02-acemoglu-restrepo-2018-modeling-automation-b","statement":"Acemoglu and Restrepo, “Modeling Automation” (2018-05) bears on M02. Argues for modelling automation as machines taking over labour-performed tasks rather than as factor-augmenting technical change.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Abstract page only (AEA website, AEA Papers and Proceedings 108, May 2018, pp. 48–53); full article not inspected. Concerns automation generally; the abstract does not name AI.","evidenceIds":["evidence-ec-m02-13"],"attributes":{"thought":{"roles":["context"],"register":"nonfiction","missing":"Automation in general; AI is not named in the abstract.","tier":{"value":1,"basis":"AEA Papers and Proceedings (core venue)."}}}},{"id":"ec-m09-korinek-stiglitz-2017-ai-income-distribution-b","statement":"Korinek and Stiglitz, “Artificial Intelligence and Its Implications for Income Distribution and Unemployment” (2017-12) bears on M09. Argues economists must say how gains from AI can be shared, not merely that they could be.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"NBER Working Paper 24174, December 2017 (NBER conference paper; later in The Economics of Artificial Intelligence). Text layer uses U+2010 hyphens. Self-critical characterisation of the profession.","evidenceIds":["evidence-ec-m09-16"],"attributes":{"thought":{"roles":["context"],"register":"nonfiction","missing":"Criticises a habitual stance of economists rather than describing the field's engagement with AI as recent or neglected.","tier":{"value":1,"basis":"NBER working paper (core venue)."}}}},{"id":"lg-oai-helen-toner-board-2023","statement":"OpenAI’s announcement of 17 November 2023 lists Helen Toner as an independent director on the board of OpenAI, Inc.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"A dated board list. It does not establish when the director joined or left; the new initial board announced on 29 November 2023 did not include her.","evidenceIds":["lg-ev-oai-helen-toner-board-2023","lg-ev-oai-helen-toner-board-2023-new-board"],"attributes":{"role":"Independent director"}},{"id":"lg-deal-mittal-inflection-cto-2024","statement":"Inflection's CEO Sean White announced on 20 May 2024 that Vibhu Mittal had joined Inflection as Chief Technology Officer.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Announcement of a joining that had taken place by the post date; the record does not establish a start date or a continuing role.","evidenceIds":["lg-ev-deal-mittal-inflection-cto-2024"],"attributes":{"role":"Chief Technology Officer"}},{"id":"lg-ant-pritchett-people-2026","statement":"Anthropic's Leadership page, as retrieved on 15 October 2026, lists Hannah Pritchett as Chief People Officer.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Shows the listing on the retrieval date only; no start date is established.","evidenceIds":["lg-ev-ant-pritchett-people-2026"],"attributes":{"role":"Chief People Officer"}},{"id":"lg-ant-mann-cofounded","statement":"An Anthropic founder slide of 2024 credits Ben Mann as a co-founder of Anthropic.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"A company-produced credit under the slide heading “Founded by top AI experts”, listing eight people as CEO, President, CSO, CTO or co-founder. Anthropic's Leadership page retrieved on 2026-10-15 does not list him among its co-founders; it is a page of current leaders, and no statement withdrawing the credit was found. The roster on either page is not established as complete.","evidenceIds":["lg-ev-ant-mann-cofounded"]},{"id":"cmp-rsp-v3-3-d3-halt","statement":"RSP · v3.3 states a provision on the comparison question “Pause, non-deployment or non-training”.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Delay commitments are tied to Appendix A competitor scenarios (see “Provisions conditioned on other developers”); outside them the edition says it would 'strongly consider' pausing (conditional wording, added in v3.1). No passage commits to not training or deploying a model without adequate safeguards (the dataset records that provision as not found in this edition). 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-v3-3-d3-halt-1","ev-cmp-rsp-v3-3-d3-halt-2"],"attributes":{"provision":{"kind":"required_action","key":"comparison-d3-halt","facet":{"id":"D3.halt","status":"stated"}}}},{"id":"cmp-meta-faf-v1-1-d3-halt","statement":"Meta FAF · v1.1 states a provision on the comparison question “Pause, non-deployment or non-training”.","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-meta-faf-v1-1-d3-halt-1","ev-cmp-meta-faf-v1-1-d3-halt-2","ev-cmp-meta-faf-v1-1-d3-halt-3"],"attributes":{"provision":{"kind":"required_action","key":"comparison-d3-halt","facet":{"id":"D3.halt","status":"stated"}}}},{"id":"ap-evan-hubinger-rlo-2019-deceptive-alignment","statement":"“Deceptive Alignment (Risks from Learned Optimization, part 4)” (2019-06-05) credits Evan Hubinger 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-evan-hubinger-rlo-2019-deceptive-alignment"]},{"id":"ap-eliezer-yudkowsky-lw-2008-fighting-a-rearguard-action-against-the","statement":"“Fighting a Rearguard Action Against the Truth” (2008-09-24) 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-fighting-a-rearguard-action-against-the"]},{"id":"cmp-rsp-v2-0-d7-internal","statement":"RSP · v2.0 states a provision on the comparison question “Internal 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-v2-0-d7-internal-1"],"attributes":{"provision":{"kind":"reporting","key":"comparison-d7-internal","facet":{"id":"D7.internal","status":"stated"}}}},{"id":"cmp-rsp-v3-0-d8-changes","statement":"RSP · v3.0 states a provision on the comparison question “Stated changes from the previous edition”.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The Introduction also says the previous RSP committed to mitigations that would reduce absolute risk “without regard to” whether other developers did the same, and names Frontier Safety Roadmaps and Risk Reports as new requirements. 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-v3-0-d8-changes-1","ev-cmp-rsp-v3-0-d8-changes-2"],"attributes":{"provision":{"kind":"revision","key":"comparison-d8-changes","facet":{"id":"D8.changes","status":"stated"}}}},{"id":"cmp-rsp-v3-4-d1-other","statement":"RSP · v3.4 states a provision on the comparison question “Other named domains”.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"A threshold row on sabotage by AI systems. Its industry-wide column calls for compartmentalization, monitoring, capability assessments and evidence on whether models tend to deceive, manipulate or sabotage their users. 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-v3-4-d1-other-1"],"attributes":{"provision":{"kind":"risk_domain","key":"comparison-d1-other","facet":{"id":"D1.other","status":"stated"}}}},{"id":"cmp-rsp-v3-4-d2-levels","statement":"RSP · v3.4 states a provision on the comparison question “Named levels or thresholds”.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The left column is headed “Capability or usage threshold” and has four rows with no ranking: non-novel chemical/biological weapons production; novel chemical/biological weapons production; “Misaligned AI systems in high-stakes settings”; automated R&D in key domains. Appendix B keeps “AI Safety Levels” (ASLs) only to refer to current levels of mitigation, not as a ladder of future capability levels. 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-v3-4-d2-levels-1"],"attributes":{"provision":{"kind":"threshold","key":"comparison-d2-levels","facet":{"id":"D2.levels","status":"stated"}}}},{"id":"cpo-ch-amd-openai-warrant","statement":"The joint AMD–OpenAI release of 6 October 2025 stated that AMD had issued OpenAI a warrant for up to 160 million shares of AMD common stock, vesting as milestones are achieved: the first tranche with the initial 1 gigawatt deployment, further tranches as purchases scale up to 6 gigawatts, and vesting further tied to AMD share-price targets and to OpenAI technical and commercial milestones.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Records the warrant as AMD describes it. It does not show that any tranche has vested or been exercised, the exercise price, or the share-price targets, which the release does not state.","evidenceIds":["evidence-cpo-ch-amd-openai-warrant-1","evidence-cpo-ch-amd-openai-warrant-2","evidence-cpo-ch-amd-openai-warrant-3"],"attributes":{"transactionStage":"announced","quantity":{"value":160000000,"unit":"shares of AMD common stock (as worded: 'up to 160 million shares')","bound":"up_to"}}},{"id":"ab-ev-gpt5-uk-aisi-jailbreak-finding","statement":"The GPT-5 System Card states that the UK AI Security Institute's safeguards team identified multiple model-level jailbreaks that overcome gpt-5-thinking's built-in refusal logic without degrading model capabilities, and that one of them evades all layers of mitigations and is being patched.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The developer's report of an external team's finding, at the date of the card. It is not the institute's own published report, says nothing about what any jailbreak could produce, and is not evidence about the state of the safeguards at any later date. No technique, prompt or output is recorded; only the existence of the finding as the card states it.","evidenceIds":["evidence-ab-ev-gpt5-uk-aisi-jailbreak-finding-1","evidence-ab-ev-gpt5-uk-aisi-jailbreak-finding-2"],"attributes":{"evaluatedSystem":"gpt-5-thinking with its deployed safeguards, as of the card","evaluation":{"what":"Robustness of the model's refusal behaviour and of the surrounding safeguard layers in the biological and chemical domain","how":"Safeguards stress testing by an external government institute with access to prototype safeguard components","result":"Multiple model-level jailbreaks identified that overcome the built-in refusal logic without degradation of capabilities; the card states the resulting content is flagged by its generation monitor, and that one jailbreak evades all layers and is being patched","caveats":"The card adds that, in practice, creating such a jailbreak would have produced numerous flags leading to account enforcement; that claim is the developer's and is not recorded as a finding."}}}],"sources":[{"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-rsp-2-0","title":"Responsible Scaling Policy — v2.0","authors":["Anthropic"],"publisher":"Anthropic","publicationDate":null,"version":"Responsible Scaling Policy v2.0; effective date 2024-10-15. v2.0 identity is supplied by the official archive and later dated changelogs; the cover is not numbered.","versionDate":"2024-10-15","canonicalUrl":"https://www-cdn.anthropic.com/616dee633636e5bd309cb73aed8622e80fe47839.pdf","evidenceUrl":"https://www-cdn.anthropic.com/616dee633636e5bd309cb73aed8622e80fe47839.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20241022154907/https://assets.anthropic.com/m/24a47b00f10301cd/original/Anthropic-Responsible-Scaling-Policy-2024-10-15.pdf","archivedAt":"2024-10-22","accessedAt":"2026-09-18","reuse":{"attribution":"Anthropic, Responsible Scaling Policy — 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-rsp-3-0","title":"Responsible Scaling Policy — v3.0","authors":["Anthropic"],"publisher":"Anthropic","publicationDate":"2026-02-24","version":"Responsible Scaling Policy v3.0; effective date 2026-02-24.","versionDate":"2026-02-24","canonicalUrl":"https://anthropic.com/responsible-scaling-policy/rsp-v3-0","evidenceUrl":"https://anthropic.com/responsible-scaling-policy/rsp-v3-0","language":"en","archivedUrl":"https://web.archive.org/web/20260224203046/https://www.anthropic.com/responsible-scaling-policy/rsp-v3-0","archivedAt":"2026-02-24","accessedAt":"2026-09-18","reuse":{"attribution":"Anthropic, Responsible Scaling Policy — v3.0.","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-rsp-3-3","title":"Responsible Scaling Policy — v3.3","authors":["Anthropic"],"publisher":"Anthropic","publicationDate":null,"version":"Responsible Scaling Policy v3.3; effective date 2026-05-26.","versionDate":"2026-05-26","canonicalUrl":"https://cdn.sanity.io/files/4zrzovbb/website/c11e84981d0a7281a1b229f3fa6af0da66eaf43f.pdf","evidenceUrl":"https://cdn.sanity.io/files/4zrzovbb/website/c11e84981d0a7281a1b229f3fa6af0da66eaf43f.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20260528170023/https://cdn.sanity.io/files/4zrzovbb/website/c11e84981d0a7281a1b229f3fa6af0da66eaf43f.pdf","archivedAt":"2026-05-28","accessedAt":"2026-09-18","reuse":{"attribution":"Anthropic, Responsible Scaling Policy — v3.3.","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-3-4","title":"Responsible Scaling Policy — v3.4","authors":["Anthropic"],"publisher":"Anthropic","publicationDate":null,"version":"Responsible Scaling Policy v3.4; effective date 2026-07-08.","versionDate":"2026-07-08","canonicalUrl":"https://cdn.sanity.io/files/4zrzovbb/website/0bacdc8440ea96e62a8766d99ebe1d4eea6d5f3a.pdf","evidenceUrl":"https://cdn.sanity.io/files/4zrzovbb/website/0bacdc8440ea96e62a8766d99ebe1d4eea6d5f3a.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20260710234036/https://cdn.sanity.io/files/4zrzovbb/website/0bacdc8440ea96e62a8766d99ebe1d4eea6d5f3a.pdf","archivedAt":"2026-07-10","accessedAt":"2026-09-18","reuse":{"attribution":"Anthropic, Responsible Scaling Policy — v3.4.","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-pf-beta","title":"Preparedness Framework — Beta","authors":["OpenAI"],"publisher":"OpenAI","publicationDate":"2023-12-18","version":"Preparedness Framework Beta; publication date 2023-12-18.","versionDate":"2023-12-18","canonicalUrl":"https://cdn.openai.com/openai-preparedness-framework-beta.pdf","evidenceUrl":"https://cdn.openai.com/openai-preparedness-framework-beta.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20231218195930/https://cdn.openai.com/openai-preparedness-framework-beta.pdf","archivedAt":"2023-12-18","accessedAt":"2026-09-18","reuse":{"attribution":"OpenAI, Preparedness Framework — Beta.","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-eval-i-pilot","title":"New report: Evaluating Language-Model Agents on Realistic Autonomous Tasks","authors":["Beth Barnes"],"publisher":"ARC Evals","publicationDate":"2023-07-31","version":"Page displays July 31, 2023 although URL contains August 1; current migrated METR-hosted expression.","canonicalUrl":"https://evals.alignment.org/blog/2023-08-01-new-report/","evidenceUrl":"https://evals.alignment.org/blog/2023-08-01-new-report/","language":"en","archivedUrl":"https://web.archive.org/web/20260616171224/https://evals.alignment.org/blog/2023-08-01-new-report/","archivedAt":"2026-06-16","accessedAt":"2026-09-27","reuse":{"attribution":"ARC Evals, New report: Evaluating Language-Model Agents on Realistic Autonomous Tasks.","licence":"No open redistribution licence asserted","note":"Attributed paraphrases and bibliographic metadata only; original documents linked, not redistributed."}},{"id":"source-eval-i-gpt4-card","title":"GPT-4 System Card","authors":["OpenAI"],"publisher":"OpenAI","publicationDate":null,"version":"60-page PDF expression retrieved 2026-09-27; printed pagination starts at 41. No publication/revision date established from this PDF. Re-inspected: 60-page PDF expression reinspected 2026-09-28; printed pagination starts at 41. Publication/revision date unknown. (Merged on 2026-10-29 with the record of the re-inspection, source-eval-ontology-i-gpt4-card, by consistency audit fix B4.)","canonicalUrl":"https://cdn.openai.com/papers/gpt-4-system-card.pdf","evidenceUrl":"https://cdn.openai.com/papers/gpt-4-system-card.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20260927044417/https://cdn.openai.com/papers/gpt-4-system-card.pdf","archivedAt":"2026-09-27","accessedAt":"2026-09-28","reuse":{"attribution":"OpenAI, GPT-4 System Card.","licence":"No open redistribution licence asserted","note":"Attributed paraphrases and bibliographic metadata only; original documents linked, not redistributed."}},{"id":"source-atlas-openai-leadership","title":"OpenAI announces leadership transition","authors":["OpenAI"],"publisher":"OpenAI","publicationDate":"2023-11-17","version":"Dated publisher page inspected 2026-09-28; current HTML, not an authenticated historical capture. Re-inspected: Web page dated November 17, 2023, inspected live 2026-10-04; quotes also matched in the Internet Archive capture of 2023-11-17 (20:30 UTC). (Merged on 2026-10-29 with the record of the re-inspection, source-openai-leadership-transition-2023, by consistency audit fix B4.)","canonicalUrl":"https://openai.com/index/openai-announces-leadership-transition/","evidenceUrl":"https://openai.com/index/openai-announces-leadership-transition/","language":"en","archivedUrl":"https://web.archive.org/web/20231117203025/https://openai.com/blog/openai-announces-leadership-transition","archivedAt":"2023-11-17","accessedAt":"2026-10-04","reuse":{"attribution":"OpenAI. OpenAI announces leadership transition.","licence":"No redistribution licence asserted.","note":"Short attributed paraphrases only; original publication is linked, not redistributed."}},{"id":"source-atlas-openai-return","title":"Sam Altman returns as CEO, OpenAI has a new initial board","authors":["Sam Altman","Bret Taylor","OpenAI"],"publisher":"OpenAI","publicationDate":"2023-11-29","version":"November 29, 2023 announcement with a separately labelled December 8 update; current HTML inspected 2026-09-28. Re-inspected: Web page dated November 29, 2023, inspected live 2026-10-04; quotes also matched in the Internet Archive capture of 2023-11-30 (02:05 UTC). (Merged on 2026-10-29 with the record of the re-inspection, source-openai-new-initial-board-2023, by consistency audit fix B4.)","canonicalUrl":"https://openai.com/index/sam-altman-returns-as-ceo-openai-has-a-new-initial-board/","evidenceUrl":"https://openai.com/index/sam-altman-returns-as-ceo-openai-has-a-new-initial-board/","language":"en","archivedUrl":"https://web.archive.org/web/20231130020514/https://openai.com/blog/sam-altman-returns-as-ceo-openai-has-a-new-initial-board","archivedAt":"2023-11-30","accessedAt":"2026-10-04","reuse":{"attribution":"OpenAI. Sam Altman returns as CEO, OpenAI has a new initial board.","licence":"No redistribution licence asserted.","note":"Short attributed paraphrases only; original publication is linked, not redistributed."}},{"id":"source-meta-faf-v1-1","title":"Frontier AI Framework — Version 1.1","authors":["Meta"],"publisher":"Meta","publicationDate":"2025-02-03","version":"Frontier AI Framework, Version 1.1, as captured by the Internet Archive on 2025-02-13, the earliest complete capture located. The original address has returned HTTP 500 since April 2026; a revised expression under the same label was captured from 2025-03-28 and is recorded separately.","versionDate":"2025-02-03","canonicalUrl":"https://ai.meta.com/static-resource/meta-frontier-ai-framework/","evidenceUrl":"https://web.archive.org/web/20250213105620/https://ai.meta.com/static-resource/meta-frontier-ai-framework/","language":"en","archivedUrl":"https://web.archive.org/web/20250213105620/https://ai.meta.com/static-resource/meta-frontier-ai-framework/","archivedAt":"2025-02-13","accessedAt":"2026-10-03","reuse":{"attribution":"Meta, Frontier AI Framework, Version 1.1 (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-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-rlo-2019-deceptive-alignment","title":"Deceptive Alignment (Risks from Learned Optimization, part 4)","authors":["Evan Hubinger","Chris van Merwijk","Vladimir Mikulik","Joar Skalse","Scott Garrabrant"],"publisher":"Alignment Forum and LessWrong","publicationDate":"2019-06-05","version":"Text checked in the post’s first stored version through the LessWrong GraphQL API on 2026-10-06.","canonicalUrl":"https://www.alignmentforum.org/posts/zthDPAjh9w6Ytbeks/deceptive-alignment","evidenceUrl":"https://www.alignmentforum.org/posts/zthDPAjh9w6Ytbeks/deceptive-alignment","language":"en","archivedUrl":"https://web.archive.org/web/20190923160252/https://www.alignmentforum.org/posts/zthDPAjh9w6Ytbeks/deceptive-alignment","archivedAt":"2019-09-23","accessedAt":"2026-10-06","reuse":{"attribution":"Evan Hubinger, Chris van Merwijk, Vladimir Mikulik, Joar Skalse, Scott Garrabrant, \"Deceptive Alignment (Risks from Learned Optimization, part 4)\", Alignment Forum and LessWrong, 2019-06-05.","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-extropians-1998-12-06-yudkowsky-2153","title":"Singularity: AI Morality (Extropians mailing list)","authors":["Eliezer Yudkowsky"],"publisher":"Extropians mailing list","publicationDate":"1998-12-06","version":"Archive HTML as served, retrieved 2026-10-07 (SHA-256 b39ca9105467de7c67ca7da085309533be010593d9be683072f57e2ca4aca564); quoted lines excluded when reading the author’s text. The served archive shows later URL rewriting and does not authenticate historical bytes. Archive at extropians.weidai.com.","canonicalUrl":"http://extropians.weidai.com/extropians.4Q98/2153.html","evidenceUrl":"http://extropians.weidai.com/extropians.4Q98/2153.html","language":"en","archivedUrl":"https://web.archive.org/web/20250525045609/https://extropians.weidai.com/extropians.4Q98/2153.html","archivedAt":"2025-05-25","accessedAt":"2026-10-07","reuse":{"attribution":"Eliezer Yudkowsky, “Singularity: AI Morality”, Extropians mailing list, 1998-12-06.","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-fighting-a-rearguard-action-against-the","title":"Fighting a Rearguard Action Against the Truth","authors":["Eliezer Yudkowsky"],"publisher":"Overcoming Bias","publicationDate":"2008-09-24","version":"Post body retrieved through the LessWrong GraphQL API on 2026-10-06 (SHA-256 ca311369027c8858532acd6d0b4da10f803e93a99b2a7c6f5ec9033c76b38bbe), content version 1.0.0. As imported to LessWrong.","canonicalUrl":"https://www.lesswrong.com/posts/PCfaLLtuxes6Jk4S2/fighting-a-rearguard-action-against-the-truth","evidenceUrl":"https://www.lesswrong.com/posts/PCfaLLtuxes6Jk4S2/fighting-a-rearguard-action-against-the-truth","language":"en","archivedUrl":"https://web.archive.org/web/20260414225554/https://www.lesswrong.com/posts/PCfaLLtuxes6Jk4S2/fighting-a-rearguard-action-against-the-truth","archivedAt":"2026-04-14","accessedAt":"2026-10-06","reuse":{"attribution":"Eliezer Yudkowsky, “Fighting a Rearguard Action Against the Truth”, Overcoming Bias, as imported to LessWrong, 2008-09-24.","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-that-tiny-note-of-discord","title":"That Tiny Note of Discord","authors":["Eliezer Yudkowsky"],"publisher":"Overcoming Bias","publicationDate":"2008-09-23","version":"Post body retrieved through the LessWrong GraphQL API on 2026-10-06 (SHA-256 ef2eeb6057de3223e07793af295329faed34e000b1f751037cd4bb813083571b), content version 1.0.0. As imported to LessWrong.","canonicalUrl":"https://www.lesswrong.com/posts/SwCwG9wZcAzQtckwx/that-tiny-note-of-discord","evidenceUrl":"https://www.lesswrong.com/posts/SwCwG9wZcAzQtckwx/that-tiny-note-of-discord","language":"en","archivedUrl":"https://web.archive.org/web/20260807050521/https://www.lesswrong.com/posts/SwCwG9wZcAzQtckwx/that-tiny-note-of-discord","archivedAt":"2026-08-07","accessedAt":"2026-10-06","reuse":{"attribution":"Eliezer Yudkowsky, “That Tiny Note of Discord”, Overcoming Bias, as imported to LessWrong, 2008-09-23.","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-0507-11477","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 1ba0e5933533210210b176aca4c4ced8e0ad4f996ad71908d77ac7e6c369aaf3); 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/0507/11477.html","evidenceUrl":"http://sl4.org/archive/0507/11477.html","language":"en","archivedUrl":"https://web.archive.org/web/20251115083834/http://sl4.org/archive/0507/11477.html","archivedAt":"2025-11-15","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-eu-ai-act-2024","title":"Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artificial intelligence (Artificial Intelligence Act)","authors":["European Parliament and Council of the European Union"],"publisher":"Official Journal of the European Union (EUR-Lex)","publicationDate":"2024-07-12","version":"Official Journal L series edition of 12.7.2024, English HTML, retrieved 2026-10-10. The text of the Internet Archive capture of 2024-07-12 matched it apart from the numbering of footnote markers in the amending articles. Later amendments (Regulation (EU) 2026/1744) are not applied to this text.","canonicalUrl":"https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=OJ:L_202401689","evidenceUrl":"https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=OJ:L_202401689","language":"en","archivedUrl":"https://web.archive.org/web/20240712193938/https://eur-lex.europa.eu/legal-content/EN/TXT/HTML/?uri=OJ:L_202401689","archivedAt":"2024-07-12","accessedAt":"2026-10-10","reuse":{"attribution":"European Parliament and Council of the European Union, Regulation (EU) 2024/1689 (Artificial Intelligence Act), Official Journal L, 12.7.2024.","licence":"Official legal text; no redistribution licence asserted","note":"Selekoa practice: metadata, attributed paraphrases and short attributed quotations under the editorial policy (at most 50 words each; at most 400 quoted words per document across the dataset, and about 10% of a document under about 2,000 words). Documents are linked externally, not redistributed. Publisher terms that forbid quotation override this practice."}},{"id":"source-acemoglu-2024-mit","title":"The Simple Macroeconomics of AI","authors":["Daron Acemoglu"],"publisher":"Paper prepared for Economic Policy","publicationDate":"2024-04-05","version":"Wayback capture of the MIT-hosted PDF (economics.mit.edu/sites/default/files/2024-04/...). Capture retrieved 2026-10-08 (SHA-256 ebe1e79ace306b02cf64753d09fb684c1311dd3391132a901ab944d10872c514). Versions: MIT-hosted paper (2024-04-05): checked; NBER Working Paper 32487 (2024-05): not compared; Economic Policy 40(121) (online 2024-08-06; issue 2025): not compared. Published version: Acemoglu, D. (2025). The simple macroeconomics of AI. Economic Policy, 40(121), 13–58. https://doi.org/10.1093/epolic/eiae042 Published online 2024-08-06 (Crossref). Working paper: NBER Working Paper No. 32487 (2024), https://doi.org/10.3386/w32487. The checked text is the MIT-hosted version dated 2024-04-05; journal wording not compared.","canonicalUrl":"https://economics.mit.edu/sites/default/files/2024-04/The%20Simple%20Macroeconomics%20of%20AI.pdf","evidenceUrl":"https://web.archive.org/web/20250106195431/https://economics.mit.edu/sites/default/files/2024-04/The%20Simple%20Macroeconomics%20of%20AI.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20250106195431/https://economics.mit.edu/sites/default/files/2024-04/The%20Simple%20Macroeconomics%20of%20AI.pdf","archivedAt":"2025-01-06","accessedAt":"2026-10-08","reuse":{"attribution":"Daron Acemoglu, “The Simple Macroeconomics of AI”, Paper prepared for Economic Policy; author-hosted PDF (MIT Economics), 2024-04-05.","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-korinek-stiglitz-2017-ai-income-distribution","title":"Artificial Intelligence and Its Implications for Income Distribution and Unemployment","authors":["Anton Korinek","Joseph E. Stiglitz"],"publisher":"NBER Working Paper 24174","publicationDate":"2017-12","version":"Full text. Current copy retrieved directly on 2026-10-09; no archive capture (Wayback unreachable), so earlier bytes are not authenticated. SHA-256 aeddf962b5c488cae68b7d48cef964de22379c78e75579c5793f0df71d5e9f40.","canonicalUrl":"https://www.nber.org/system/files/working_papers/w24174/w24174.pdf","evidenceUrl":"https://www.nber.org/system/files/working_papers/w24174/w24174.pdf","language":"en","accessedAt":"2026-10-09","reuse":{"attribution":"Anton Korinek, Joseph E. Stiglitz, “Artificial Intelligence and Its Implications for Income Distribution and Unemployment”, NBER Working Paper 24174, 2017-12.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"source-acemoglu-restrepo-2018-modeling-automation","title":"Modeling Automation","authors":["Daron Acemoglu","Pascual Restrepo"],"publisher":"AEA Papers and Proceedings 108, 2018, pp. 48–53","publicationDate":"2018-05","version":"Abstract page only. Current copy retrieved directly on 2026-10-09; no archive capture (Wayback unreachable), so earlier bytes are not authenticated. SHA-256 44b7536ed812f0aa4db76e82aee9012fdb8cfad14764c580fb2f9dc79cbd39f0. Full text not retrieved; only the AEA abstract page. Acemoglu, D., & Restrepo, P. (2018). Modeling automation. AEA Papers and Proceedings, 108, 48–53. https://doi.org/10.1257/pandp.20181020","canonicalUrl":"https://www.aeaweb.org/articles?id=10.1257/pandp.20181020","evidenceUrl":"https://www.aeaweb.org/articles?id=10.1257/pandp.20181020","language":"en","accessedAt":"2026-10-09","reuse":{"attribution":"Daron Acemoglu, Pascual Restrepo, “Modeling Automation”, AEA Papers and Proceedings 108, 2018, pp. 48–53, 2018-05.","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":"lg-src-ant-leadership-2026","title":"Leadership at Anthropic (web page, undated; retrieved 2026-10-15)","authors":["Anthropic"],"publisher":"Anthropic","publicationDate":null,"version":"Undated page, retrieved on 2026-10-15 (SHA-256 a2bc8182bd18aff065d4bce72994c832816f98e4f099d956173e849b7baa789c) directly from the publisher. Its wording can change; earlier wording is not authenticated.","canonicalUrl":"https://www.anthropic.com/company/leadership","evidenceUrl":"https://www.anthropic.com/company/leadership","language":"en","archivedUrl":"https://web.archive.org/web/20261015210858/https://www.anthropic.com/company/leadership","archivedAt":"2026-10-15","accessedAt":"2026-10-15","reuse":{"attribution":"Anthropic, “Leadership at Anthropic (web page, undated; retrieved 2026-10-15)”.","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-deal-inflection-exec-2024","title":"Redefining the Future of AI","authors":["Sean White"],"publisher":"Inflection AI","publicationDate":"2024-05-20","version":"Dated post, retrieved on 2026-10-15 (SHA-256 00da8e8a510757e221dfa7959ae036c2335558a258e839766f54fe514a1475da) directly from the publisher.","canonicalUrl":"https://inflection.ai/blog/redefining-the-future-of-ai","evidenceUrl":"https://inflection.ai/blog/redefining-the-future-of-ai","language":"en","archivedUrl":"https://web.archive.org/web/20260901215817/https://inflection.ai/blog/redefining-the-future-of-ai","archivedAt":"2026-09-01","accessedAt":"2026-10-15","reuse":{"attribution":"Sean White, “Redefining the Future of AI”, 2024-05-20.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"source-cpo-ch-openai-amd-2025","title":"AMD and OpenAI Announce Strategic Partnership to Deploy 6 Gigawatts of AMD GPUs","authors":["AMD (Investor Relations)"],"publisher":"AMD (Investor Relations)","publicationDate":"2025-10-06","version":"Retrieved for the candidate pack on 2026-10-28 (SHA-256 ba034c7b34d3bc716b7f26832720c7b2c69434f2b15b0d5634b0e50924ce589e). Press release (GLOBE NEWSWIRE) on ir.amd.com as served on 2026-10-28; dated October 06, 2025, 7:00 am EDT. OpenAI's own post on the same announcement could not be retrieved (openai.com refuses direct retrieval from this environment); the cross-family checker should open it or its capture. Retrieved again at admission on 2026-10-30 (SHA-256 ba034c7b34d3bc716b7f26832720c7b2c69434f2b15b0d5634b0e50924ce589e, the same bytes). Its wording can change; earlier wording is not authenticated.","canonicalUrl":"https://ir.amd.com/news-events/press-releases/detail/1260/amd-and-openai-announce-strategic-partnership-to-deploy-6-gigawatts-of-amd-gpus","evidenceUrl":"https://ir.amd.com/news-events/press-releases/detail/1260/amd-and-openai-announce-strategic-partnership-to-deploy-6-gigawatts-of-amd-gpus","language":"en","accessedAt":"2026-10-28","reuse":{"attribution":"AMD (Investor Relations), “AMD and OpenAI Announce Strategic Partnership to Deploy 6 Gigawatts of AMD GPUs”, 2025-10-06.","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-gpt5-card","title":"GPT-5 System Card","authors":["OpenAI"],"publisher":"OpenAI","publicationDate":null,"version":"Retrieved for the candidate pack on 2026-10-28 (SHA-256 f9f9c72d08cba81dcbb189129119357c8c32eca4e0b16c813aca386aed6b0459). PDF dated August 13, 2025 on its cover, as served on 2026-10-28. Retrieved again at admission on 2026-10-30 (SHA-256 f9f9c72d08cba81dcbb189129119357c8c32eca4e0b16c813aca386aed6b0459, the same bytes).","canonicalUrl":"https://cdn.openai.com/gpt-5-system-card.pdf","evidenceUrl":"https://cdn.openai.com/gpt-5-system-card.pdf","language":"en","accessedAt":"2026-10-28","reuse":{"attribution":"OpenAI, “GPT-5 System Card”.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}}],"evidence":[{"id":"evidence-rsp-v3-4-coverage-date","sourceRole":"primary","support":"supports","locator":"PDF p. 21, Changelog, July 8, 2026 (RSP v3.4) entry, item 3, first sentence","passage":"It adjusts the scope of Risk Reports to cover Anthropic’s models and activities as of a given coverage date up to 30 days prior to that report’s publication.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-03","note":"Verbatim excerpt checked by an automated research assistant on 2026-10-03 against the inspected PDF, byte-identical to the Internet Archive capture of 2026-07-10; the PDF text layer omits the word spaces around \"coverage date\", restored here; it replaces an earlier attributed paraphrase. Human editorial review is pending."},{"id":"evidence-eval-i-openai-gpt4-card","sourceRole":"primary","support":"supports","locator":"PDF page 1, title and OpenAI author credit.","passage":"OpenAI is the credited author of the system card.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-09-27","note":"Original publisher passage machine-inspected. This is not human review or independent reproduction of the reported activity."},{"id":"evidence-eval-i-arc-pilot","sourceRole":"primary","support":"supports","locator":"Displayed date; Background and Approach sections.","passage":"ARC Evals describes its pilot evaluations of Claude- and GPT-4-based agents.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-09-27","note":"Original publisher passage machine-inspected. This is not human review or independent reproduction of the reported activity."},{"id":"evidence-prov-preparedness-beta-2023-safety-enhancing-development-exception","sourceRole":"primary","support":"supports","locator":"PDF p. 21 (printed p. 21), Governance › Safety baselines, “Restricting development”, third sentence","passage":"Note that this should not preclude safety-enhancing development.","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 2023-12-18. Extraction note: “this” refers to the development restriction stated in the two preceding sentences of the same paragraph. Human editorial review is pending."},{"id":"evidence-ts-b1-yudkowsky-1996-12-31-5142-context","sourceRole":"primary","support":"supports","locator":"Message body, fourth unquoted paragraph","passage":"less likely to let a friendly Power out than an evil one, because the evil one could lie.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-06","note":"The same passage as the B5 record."},{"id":"evidence-yk-t02-e","sourceRole":"primary","support":"supports","locator":"That Tiny Note of Discord, opening paragraph","passage":"When we last left Eliezer1997, he believed that any superintelligence would automatically do what was \"right\", and indeed would understand that better than we could … (The flaw being the unconsidered and unjustified equation of \"universally compelling argument\" with \"right\".)","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-06","note":"Candidate pack passage T02-e. The quotation (40 words) matched the retrieved expression (SHA-256 ef2eeb6057de3223e07793af295329faed34e000b1f751037cd4bb813083571b) when checked on 2026-10-06."},{"id":"evidence-yk-t04-e1","sourceRole":"primary","support":"supports","locator":"message body, paragraph beginning 'Once again, we have a'","passage":"Do you really know all the logical consequences of placing a large value on human survival? Would you care to define \"human\" for me? Oops! Thanks to your overly rigid definition, you will live for billions and trillions and googolplexes of years","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-07","note":"Candidate pack passage T04-e1. The quotation (42 words) matched the retrieved expression (SHA-256 b39ca9105467de7c67ca7da085309533be010593d9be683072f57e2ca4aca564) when checked on 2026-10-07."},{"id":"evidence-yk-t08-s6","sourceRole":"primary","support":"supports","locator":"message body, paragraph beginning 'In truth it seems pointless to me anyway'","passage":"I do not expect to see an organized governmental agency that even understands the concept of UFAI … this side of the Singularity.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-07","note":"Candidate pack passage T08-s6. The quotation (23 words) matched the retrieved expression (SHA-256 1ba0e5933533210210b176aca4c4ced8e0ad4f996ad71908d77ac7e6c369aaf3) when checked on 2026-10-07."},{"id":"evidence-aia-art111-earlier-models-1","sourceRole":"primary","support":"supports","locator":"Article 111(3)","passage":"shall take the necessary steps in order to comply with the obligations laid down in this Regulation by 2 August 2027","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-10","note":"Verbatim quote, matched mechanically (after whitespace normalisation) by the proposing session against the text retrieved on 2026-10-10. Checked by fresh-context machine audit on 2026-10-10 (review/frontier-governance-audit-2026-10-10.md). Human editorial review is pending."},{"id":"evidence-aia-art111-earlier-models-2","sourceRole":"primary","support":"supports","locator":"Article 111(3), opening","passage":"The paragraph applies to providers of general-purpose AI models that have been placed on the market before 2 August 2025.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-10","note":"Editorial paraphrase proposed by the proposing session from the text retrieved on 2026-10-10. Checked by fresh-context machine audit on 2026-10-10 (review/frontier-governance-audit-2026-10-10.md). Human editorial review is pending."},{"id":"evidence-ec-g09-i","sourceRole":"primary","support":"supports","locator":"Section 3 \"A Preliminary Quantitative Evaluation\", paragraph giving the 0.71% estimate","passage":"This is a nontrivial, but modest effect, and certainly much less than both the revolutionary changes some are predicting","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-14","note":"Candidate pack passage G09-i (research/economics-growth). The quotation (19 words) matched the checked expression (acemoglu-mit) in the pack's verification. Re-checked at admission on 2026-10-14 against a copy retrieved that day.","contextOnly":{"reason":"Alone, it reads as a rebuttal of explosive-growth forecasts; the estimate covers ten years and existing AI.","requires":"The ten-year horizon and the task-exposure assumptions (G08-a)."}},{"id":"evidence-ec-m02-13","sourceRole":"primary","support":"supports","locator":"AEA article page, abstract","passage":"modeling it as the process of machines replacing tasks previously performed by labor is both descriptively realistic","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-14","note":"Candidate pack passage M02-13 (research/economics-mainstream). The quotation (17 words) matched the retrieved current copy (SHA-256 44b7536ed812f0aa4db76e82aee9012fdb8cfad14764c580fb2f9dc79cbd39f0) in the pack's verification. Re-checked at admission on 2026-10-14 against a copy retrieved that day, whose SHA-256 differs from the recorded one.","contextOnly":{"reason":"Fresh-context audit (A3) judged that the quotation misleads when shown alone.","requires":"Quote truncated mid-sentence: continues 'and leads to distinct and plausible predictions'."}},{"id":"evidence-ec-m09-16","sourceRole":"primary","support":"supports","locator":"pdf page 4; section 1 Introduction","passage":"We believe that the primary economic challenge posed by the proliferation of AI will be one of income distribution. We economists set ourselves too easy a goal if we just say that technological progress can make everybody better off","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-14","note":"Candidate pack passage M09-16 (research/economics-mainstream). The quotation (39 words) matched the retrieved current copy (SHA-256 aeddf962b5c488cae68b7d48cef964de22379c78e75579c5793f0df71d5e9f40) in the pack's verification. Re-checked at admission on 2026-10-14 against a copy retrieved that day."},{"id":"lg-ev-oai-helen-toner-board-2023","sourceRole":"primary","support":"supports","locator":"Paragraph beginning “OpenAI’s board of directors consists of”","passage":"The board list names Helen Toner as one of three independent directors; chief scientist Ilya Sutskever is the fourth board member listed.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The paraphrase’s verbatim anchor “Emerging Technology’s Helen Toner” matched the retrieved text on 2026-10-15 and again in a fresh retrieval the same day. Read through the third-party reader relay r.jina.ai, which renders the publisher’s current page (the publisher refused direct retrieval); not an archive copy. Matching ignored differences of whitespace, case, quotation marks, apostrophes, dashes and Unicode compatibility forms, so the source’s typography can differ. The paraphrase is Selekoa’s; only the anchor is the source’s wording. No human editorial review.","retrievalRoute":"reader_relay"},{"id":"lg-ev-oai-helen-toner-board-2023-new-board","sourceRole":"primary","support":"context","locator":"Altman’s message, opening paragraph (29 November 2023)","passage":"The new initial board will consist of Bret Taylor (Chair), Larry Summers, and Adam D’Angelo.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The quotation (15 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-deal-mittal-inflection-cto-2024","sourceRole":"primary","support":"supports","locator":"Paragraph 4","passage":"the first three members of my new executive team have joined Inflection - Vibhu Mittal as Chief Technology Officer","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The quotation (19 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-ant-pritchett-people-2026","sourceRole":"primary","support":"supports","locator":"Section “Executives”, Hannah Pritchett entry","passage":"Hannah Pritchett's entry gives the title Chief People Officer.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The paraphrase’s verbatim anchor “Chief People Officer” matched the retrieved text on 2026-10-15 and again in a fresh retrieval the same day. Read directly from the publisher. Matching ignored differences of whitespace, case, quotation marks, apostrophes, dashes and Unicode compatibility forms, so the source’s typography can differ. The paraphrase is Selekoa’s; only the anchor is the source’s wording. No human editorial review.","retrievalRoute":"direct"},{"id":"lg-ev-ant-mann-cofounded","sourceRole":"primary","support":"supports","locator":"PDF page 6, “Our business”, founder panel, Ben Mann row (row layout checked on the rendered page)","passage":"The Ben Mann row reads “Co-founder / Former GPT-3 and API Engineer at OpenAI”.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The paraphrase’s verbatim anchor “Ben Mann” matched the retrieved text on 2026-10-15 and again in a fresh retrieval the same day. Read directly from the publisher. Matching ignored differences of whitespace, case, quotation marks, apostrophes, dashes and Unicode compatibility forms, so the source’s typography can differ. The paraphrase is Selekoa’s; only the anchor is the source’s wording. No human editorial review.","retrievalRoute":"direct"},{"id":"ev-cmp-rsp-v3-3-d3-halt-1","sourceRole":"primary","support":"supports","locator":"PDF p. 17 (printed p. 16), Appendix A: Commitments Related to Competitors, table row 'Anthropic in the lead', Commitment column","passage":"We will require a strong argument that catastrophic risk is contained, along the lines of our recommendations for industry-wide safety (see Section 1). We will delay AI development and deployment as needed to achieve this, until and unless we no longer believe we have a significant lead.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-27","note":"Existing dataset passage for this edition, which contains the comparison’s shorter quotation, first checked on 2026-10-03. The comparison’s quotation was 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. Applies only in Appendix A's 'Anthropic in the lead' scenario. 'This' refers to a strong argument that catastrophic risk is contained, along the lines of the industry-wide recommendations. Human editorial review is pending."},{"id":"ev-cmp-rsp-v3-3-d3-halt-2","sourceRole":"primary","support":"supports","locator":"PDF p. 17 (printed p. 16), Appendix A: Commitments Related to Competitors, introductory paragraph, second-to-last sentence","passage":"the commitments below do not preclude us from taking cautionary action, such as refraining from training or deploying models, in other circumstances.","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 next sentence says Anthropic 'would strongly consider pausing development and/or deployment' to improve the safety profiles of its models. Human editorial review is pending."},{"id":"ev-cmp-meta-faf-v1-1-d3-halt-1","sourceRole":"primary","support":"supports","locator":"PDF p. 12 (printed p. 12), §3.3 Risk thresholds, first paragraph, third sentence","passage":"If a frontier AI is assessed to have reached the critical risk threshold and cannot be mitigated, we will stop development and implement the measures outlined in Table 1.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-27","note":"Passage recorded in the dataset for this edition, checked by an automated research assistant on 2026-10-03 against the Internet Archive capture of 2025-02-13, and matched against the same capture on 2026-10-27 by a reviewer from a different model family, which retrieved and hashed it (decoded PDF SHA-256 9fa301df…cc21; research/frontier-framework-comparison/codex-meta-v1-1-2026-10-27.md). Human editorial review is pending."},{"id":"ev-cmp-meta-faf-v1-1-d3-halt-2","sourceRole":"primary","support":"supports","locator":"PDF p. 17 (printed p. 17), §4.2 Evaluation and mitigation, paragraph beginning 'We may take into account monetary costs', second sentence","passage":"If the results of our evaluations indicate that a frontier AI has a “high” risk threshold by providing significant uplift towards realization of a catastrophic outcome we will not release the frontier AI externally.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-27","note":"Passage recorded in the dataset for this edition, checked by an automated research assistant on 2026-10-03 against the Internet Archive capture of 2025-02-13, and matched against the same capture on 2026-10-27 by a reviewer from a different model family, which retrieved and hashed it (decoded PDF SHA-256 9fa301df…cc21; research/frontier-framework-comparison/codex-meta-v1-1-2026-10-27.md). Human editorial review is pending."},{"id":"ev-cmp-meta-faf-v1-1-d3-halt-3","sourceRole":"primary","support":"supports","locator":"PDF p. 13 (printed p. 13), §3.3 Risk thresholds, Table 1: Risk thresholds for frontier AI, Critical row, Measures column, first bullet, second sentence","passage":"If a model appears to uniquely enable the execution of a threat scenario we will pause development while we investigate whether barriers to realizing the catastrophic outcome remain.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-27","note":"Passage recorded in the dataset for this edition, checked by an automated research assistant on 2026-10-03 against the Internet Archive capture of 2025-02-13, and matched against the same capture on 2026-10-27 by a reviewer from a different model family, which retrieved and hashed it (decoded PDF SHA-256 9fa301df…cc21; research/frontier-framework-comparison/codex-meta-v1-1-2026-10-27.md). Human editorial review is pending."},{"id":"evidence-ap-evan-hubinger-rlo-2019-deceptive-alignment","sourceRole":"primary","support":"supports","locator":"“by” line beneath the title; the current page displays the username “evhub” in its place","passage":"Evan Hubinger","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 Internet Archive capture of 2019-09-23; the preparer’s matcher had not found the name in a byline position. Human editorial review is pending.","retrievalRoute":"direct"},{"id":"evidence-ap-eliezer-yudkowsky-lw-2008-fighting-a-rearguard-action-against-the","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 and checked by a reviewer from a different model family on 2026-10-27. Human editorial review is pending.","retrievalRoute":"direct"},{"id":"ev-cmp-rsp-v2-0-d7-internal-1","sourceRole":"primary","support":"supports","locator":"PDF p. 16 (printed p. 12), §7.1 Internal Governance, numbered item 3 ('Transparency'), first sentence","passage":"The edition says Anthropic will share summaries of its Capability Reports and Safeguards Reports with regular-clearance staff, with highly sensitive information redacted.","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. The same section says the Responsible Scaling Officer notifies the Board of noncompliance posing material risk and updates it regularly on noncompliance (§7.1 items 1 and 5); decisions and reports go to the Board and LTBT (§4.3). Human editorial review is pending."},{"id":"ev-cmp-rsp-v3-0-d8-changes-1","sourceRole":"primary","support":"supports","locator":"PDF p. 19, Changelog, entry 'February 24, 2026 (RSP v3.0)'","passage":"The v3.0 changelog entry describes the update as a comprehensive rewrite of the RSP.","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. The entry points to a separate summary of changes that is not part of this document. Human editorial review is pending."},{"id":"ev-cmp-rsp-v3-0-d8-changes-2","sourceRole":"primary","support":"supports","locator":"PDF p. 3, Introduction, paragraph beginning 'We now separate'","passage":"The Introduction says the edition now separates the plans Anthropic expects to achieve as a company, whatever other companies do, from its more ambitious recommendations for the whole industry.","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-rsp-v3-4-d1-other-1","sourceRole":"primary","support":"supports","locator":"PDF p. 7, §1, threshold table, row title, left column (“Capability or usage threshold”)","passage":"The edition’s threshold table has a row titled “Misaligned AI systems in high-stakes settings”.","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. The row concerns AI systems that are heavily relied on, have wide access to sensitive assets and a moderate ability to act autonomously toward goals and to use subterfuge, such that they could plausibly carry out sabotage that irreversibly and substantially raises the chance of a global catastrophe later on. Human editorial review is pending."},{"id":"ev-cmp-rsp-v3-4-d2-levels-1","sourceRole":"primary","support":"supports","locator":"PDF p. 4, §1 Our Recommendations for Industry-Wide Safety, first paragraph","passage":"The edition says the left column of its three-column table sets out the capability thresholds at which heightened mitigations would be needed.","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":"evidence-cpo-ch-amd-openai-warrant-1","sourceRole":"primary","support":"supports","locator":"body, paragraph 4","passage":"AMD has issued OpenAI a warrant for up to 160 million shares of AMD common stock, structured to vest as specific milestones are achieved.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-30","note":"Candidate pack passage, compute, power and orbit (research/compute-power-orbit). The quotation (24 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 (2026-10-29/30, item 63) answered the record’s question blind from fresh retrievals and agreed with the record. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-cpo-ch-amd-openai-warrant-2","sourceRole":"primary","support":"supports","locator":"body, paragraph 4","passage":"The first tranche vests with the initial 1 gigawatt deployment, with additional tranches vesting as purchases scale up to 6 gigawatts.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-30","note":"Candidate pack passage, compute, power and orbit (research/compute-power-orbit). The quotation (21 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 (2026-10-29/30, item 63) answered the record’s question blind from fresh retrievals and agreed with the record. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-cpo-ch-amd-openai-warrant-3","sourceRole":"primary","support":"supports","locator":"body, paragraph 4","passage":"Vesting is further tied to AMD achieving certain share-price targets and to OpenAI achieving the technical and commercial milestones","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-30","note":"Candidate pack passage, compute, power and orbit (research/compute-power-orbit). The quotation (19 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 (2026-10-29/30, item 63) answered the record’s question blind from fresh retrievals and agreed with the record. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-ab-ev-gpt5-uk-aisi-jailbreak-finding-1","sourceRole":"primary","support":"supports","locator":"PDF p. 53 (printed p. 52), §5.3.3.5 External government red teaming","passage":"The UK AISI's Safeguards team identified multiple model-level jailbreaks that overcome gpt-5-thinking's built-in refusal logic without degradation of model capabilities","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-30","note":"Candidate pack passage, AI and biosecurity (research/ai-bio). The quotation (20 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 38) re-read the record’s sources and found full support. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-ab-ev-gpt5-uk-aisi-jailbreak-finding-2","sourceRole":"primary","support":"supports","locator":"PDF p. 53 (printed p. 52), §5.3.3.5 External government red teaming, sentence following the jailbreak finding","passage":"One of the jailbreaks evades all layers of mitigations and is being patched.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-30","note":"Candidate pack passage, AI and biosecurity (research/ai-bio). The quotation (13 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. Added after the audit of 2026-10-28: the statement's second half rested on no cited passage. 'Is being patched' is the card's own present tense at its date of publication; the record asserts nothing about any later state. A cross-family machine check (2026-10-30, item 38) re-read the record’s sources and found full support. No human editorial review.","retrievalRoute":"direct"}],"at":{"entities":[0,1,12,19,42,45,48,49,53,112,113,199,207,235,295,331,342,351,353,357,405,421,425,481,491,512,514,522,534,540,559,638,680,681,815,886,1131,1245],"claims":[94,263,264,449,884,998,1016,1047,1209,1292,1325,1461,1666,1690,1911,1931,2085,2150,2265,2443,2501,2540,2581,2582,3209,3395],"sources":[17,36,39,42,43,48,113,114,124,125,178,297,336,372,388,392,448,457,485,492,512,569,597,947,1080],"evidence":[111,284,285,481,1099,1224,1243,1275,1480,1481,1619,1656,1821,2097,2098,2136,2473,2495,2696,2697,2790,2791,2792,2912,3090,3148,3191,3192,3239,3240,4362,4363,4364,4667,4668]},"version":"1lg47t7"}