{"entities":[{"id":"anthropic","name":"Anthropic"},{"id":"google-deepmind","name":"Google DeepMind"},{"id":"xai","name":"xAI"},{"id":"amazon","name":"Amazon"},{"id":"eliezer-yudkowsky","name":"Eliezer Yudkowsky"},{"id":"fsf-2024","name":"Frontier Safety Framework (2024)"},{"id":"fsf-v2-0","name":"Google DeepMind Frontier Safety Framework — v2.0"},{"id":"faif-2025","name":"xAI Frontier Artificial Intelligence Framework — December 2025 edition"},{"id":"faif-2026","name":"xAI Frontier Artificial Intelligence Framework — June 2026 edition"},{"id":"softbank-group","name":"SoftBank Group"},{"id":"project-rainier","name":"Project Rainier"},{"id":"meta","name":"Meta"},{"id":"tsinghua","name":"Tsinghua University / 清华大学"},{"id":"tsinghua-keg","name":"Tsinghua Knowledge Engineering Group / 知识工程实验室"},{"id":"minsky-1961-steps","name":"Steps Toward Artificial Intelligence"},{"id":"drexler-1986-engines-of-creation","name":"Engines of Creation: The Coming Era of Nanotechnology"},{"id":"christiano-2018-takeoff-speeds","name":"Takeoff speeds"},{"id":"christiano-2022-where-i-agree-and-disagree","name":"Where I agree and disagree with Eliezer"},{"id":"fast-or-hard-takeoff","name":"Fast or hard takeoff"},{"id":"designed-deception-of-testers","name":"Systems designed to mislead their testers"},{"id":"feedback-instils-values","name":"Training on human feedback instils human values"},{"id":"ai-solves-alignment","name":"AI systems will solve alignment for us"},{"id":"time-2023-shut-it-all-down","name":"Pausing AI Developments Isn't Enough. We Need to Shut it All Down"},{"id":"extropians-1999-08-04-yudkowsky-1587","name":"Re: Does a parental model work for having A.I. that can be trusted?"},{"id":"lw-2007-thou-art-godshatter","name":"Thou Art Godshatter"},{"id":"lw-2008-what-i-think-if-not-why","name":"What I Think, If Not Why"},{"id":"eu-ai-act-2024","name":"Regulation (EU) 2024/1689 (Artificial Intelligence Act), as published in the Official Journal"},{"id":"accelerating-growth-from-automated-research","name":"Automated research produces accelerating growth"},{"id":"task-based-exposure-measurement","name":"Exposure to automation measured from task content"},{"id":"hanson-economic-growth-given-machine-intelligence","name":"Economic Growth Given Machine Intelligence"},{"id":"frey-osborne-2013-future-of-employment","name":"The Future of Employment: How Susceptible Are Jobs to Computerisation?"},{"id":"ai-improving-ai-accelerates-capability","name":"Machine intelligence improving itself accelerates its capability growth"},{"id":"trammell-korinek-2026-growth-under-transformative-ai-nber-revision","name":"Economic Growth under Transformative AI"},{"id":"ahmad-al-dahle","name":"Ahmad Al-Dahle"},{"id":"anca-dragan","name":"Anca Dragan"},{"id":"bytedance-seed","name":"ByteDance Seed / 字节跳动 Seed"},{"id":"yuhuai-wu","name":"Yuhuai (Tony) Wu"},{"id":"marvin-minsky","name":"Marvin Minsky"},{"id":"seed-coder-licence","name":"Seed-Coder-8B-Instruct licence statement, revision 1c5af72a"},{"id":"tc260","name":"National Technical Committee 260 on Cybersecurity of SAC (全国网络安全标准化技术委员会)"},{"id":"tc260-asgf-v3-0","name":"AI Safety Governance Framework 3.0 (人工智能安全治理框架3.0)"},{"id":"bletchley-chair-safety-testing-2023","name":"Safety Testing: Chair's Statement of Session Outcomes, 2 November 2023"},{"id":"us-uk-aisi-preeval-claude-35-sonnet-2024","name":"Pre-Deployment Evaluation of Anthropic’s Upgraded Claude 3.5 Sonnet (US AISI and UK AISI)"},{"id":"aws-us-government-ai-expansion","name":"AWS AI and HPC expansion across Top Secret, Secret and GovCloud (US) Regions"},{"id":"stargate-milam-county","name":"Stargate site, Milam County, Texas"},{"id":"anthropic-transparency-hub-2026-10-28","name":"Transparency hub (Anthropic), as served on 28 October 2026"},{"id":"claude-sonnet-4-5-system-card","name":"System Card: Claude Sonnet 4.5"},{"id":"bio-misuse-capability-evaluation","name":"Biological misuse capability evaluation"}],"claims":[{"id":"anthropic-rainier-2025","statement":"Amazon reported launching Project Rainier to build and deploy Anthropic’s Claude models.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Company-reported deployed workload infrastructure; not independent performance validation or Anthropic ownership. The cited passage states the cluster’s launch and purpose; it does not state Anthropic’s use at this date. Amazon’s 2026-04-20 page states active use (later snapshot).","evidenceIds":["evidence-anthropic-rainier-2025"],"attributes":{"transactionType":"infrastructure_use","relationshipContext":"external","transactionStage":"deployed"}},{"id":"cn-keg-tsinghua","statement":"Tsinghua’s 2021 account identifies KEG as its computer-science research laboratory.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Internal academic laboratory, distinct from the separately named company Zhipu.","evidenceIds":["evidence-cn-keg-tsinghua"]},{"id":"prov-fsf-v2-0-government-disclosure","statement":"FSF v2.0 states that if Google DeepMind assesses that a model has reached a CCL that poses an unmitigated and material risk to overall public safety, it aims to share information with appropriate government authorities where it will facilitate the development of safe AI.","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-fsf-v2-0-government-disclosure"],"attributes":{"provision":{"kind":"decision_authority","key":"fsf-government-disclosure","parties":[{"role":"decides","actor":"Google DeepMind"},{"role":"informed","actor":"Appropriate government authorities"}]}}},{"id":"ts-c2-drexler-1986","statement":"Engines of Creation’s note on social AI says a Turing-test passer would of necessity be designed to understand and deceive people.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Recorded under C2 because the deception is by design; the wording is attested from 1997.","evidenceIds":["evidence-ts-c2-drexler-1986"],"attributes":{"thought":{"roles":["articulation"],"register":"nonfiction","attested":{"date":"1997-07-09","note":"Wording checked in the Foresight Institute’s web edition (© 1986 K. Eric Drexler; published for the web by Russell Whitaker), whose earliest captures are of 1997-07-09; a reader’s message of October 1996 on the site shows the edition was online by then. Not compared with the 1986 printing."}}}},{"id":"ts-a4-christiano-2018-takeoff-contests","statement":"Christiano’s post operationalises slow takeoff as a complete four-year interval in which world output doubles before the first one-year doubling, argues that it is more likely, and holds that recursive self-improvement, though it will lead to runaway growth, is not on its own an argument for a discontinuity.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"A claim about the speed of the transition; the post says an intelligence explosion \"seems very likely\" and that it is not about whether one occurs.","evidenceIds":["evidence-ts-a4-christiano-2018-takeoff-contests","evidence-ts-a4-christiano-2018-takeoff-contests-2"],"attributes":{"thought":{"roles":["contesting"],"register":"nonfiction","qualifier":"Argues for slow takeoff and against a discontinuity.","engagement":{"kind":"engages-premise","rationale":"Gives a reason against the premise of a short transition: before AI is great at self-improvement it will be mediocre at it, so runaway growth need not be a discontinuity."}}}},{"id":"ts-a4-christiano-2022-context","statement":"Christiano’s post disputes the picture of AI systems that hide their capabilities while lying in wait, arguing that SGD will aggressively select against systems that do not do impressive-looking things.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"A conditional disagreement about a capability leap after concealment.","evidenceIds":["evidence-ts-a4-christiano-2022-context"],"attributes":{"thought":{"roles":["context"],"register":"nonfiction","missing":"It disputes a capability leap revealed after concealment, arguing that training exposes less capable precursors first; it states no period for the transition."}}},{"id":"yk-y05-lw-2007-thou-art-godshatter-b","statement":"The post “Thou Art Godshatter” (2007-11-13) bears on Y05. Discussing human evolution, he observes that selection for reproductive fitness produced many separate desires rather than a desire for fitness itself; he later applies the analogy to AI training (AGI Ruin item 16).","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Premise addressed: Optimising a system for human approval or reward produces internalised human values. Imported Overcoming Bias post (2007-11-13). About evolution, not AI; included as the earliest located statement of the outer/inner analogy he later applies to training. A 2007 passage about evolution and human values; reading it as a comment on training AI from human feedback projects a later analogy (AGI Ruin item 16) back onto it.","evidenceIds":["evidence-yk-t05-c"],"attributes":{"thought":{"roles":["context"],"register":"nonfiction","missing":"About evolution and human desires, not about training AI on human feedback."}}},{"id":"yk-y05-extropians-1999-08-04-yudkowsky-1587-c","statement":"The Extropians message “Re: Does a parental model work for having A.I. that can be trusted?” (1999-08-04) contests Y05. In the thread 'Does a parental model work for having A.I. that can be trusted?', he calls the parenting model anthropomorphic: whether AIs are raised permissively or harshly makes no difference, because parenting relies on human emotions an AI will not have.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Premise addressed: AI training will instil values the way upbringing instils morals in children. Concerns child-rearing as a model for AI motivation, before machine-learning feedback training was the method at issue. A near-identical copy was posted seconds apart (a second Extropians message of 1999-08-04, 'isn't going to exist'). Quotation shortened for the ~10% cap. Quotation shortened to respect the ~10% cap for a short message.","evidenceIds":["evidence-yk-t05-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 AI training will instil values the way upbringing instils morals in children."}}}},{"id":"yk-y07-time-2023-shut-it-all-down-c","statement":"“Pausing AI Developments Isn't Enough. We Need to Shut it All Down” (2023-03-29) contests Y07. In an op-ed arguing that a six-month moratorium is insufficient, he cites a leading lab's stated intention to have future AI do alignment work as evidence that there is no viable survival plan.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Premise addressed: An AI weak enough to be safe is strong enough to solve alignment. Characterises a named lab's published plan in his own words; it records his reaction, not an analysis of that plan.","evidenceIds":["evidence-yk-t07-e"],"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 an AI weak enough to be safe is strong enough to solve alignment, even when read with its message or dialogue turn."}}}},{"id":"aia-art101-gpai-fines","statement":"Article 101(1) provides that the Commission may impose on providers of general-purpose AI models fines not exceeding 3% of their annual total worldwide turnover in the preceding financial year or EUR 15,000,000, whichever is higher, when it finds that the provider intentionally or negligently (a) infringed the relevant provisions, (b) failed to comply with a request for a document or information or supplied incorrect, incomplete or misleading information, (c) failed to comply with a measure requested under Article 93, or (d) failed to give access to the model for an evaluation under Article 92.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"A ceiling and a power, not a record of any fine. No fine under this Article is recorded. Article 113, third paragraph, point (b), excepts Article 101 (in Chapter XII) from the 2 August 2025 date of application (separate record). It states what the Regulation requires of the categories of provider it defines. Whether any named developer’s model falls within a category, and whether any developer complies, is legal application and is not recorded. A statutory duty is not evidence of implementation or enforcement. Regulation (EU) 2026/1744 (published 24 July 2026) amends other articles of the AI Act. Its text, inspected on 2026-10-10, does not list this article among those it amends; the consolidated AI Act was not inspected.","evidenceIds":["evidence-aia-art101-gpai-fines-1","evidence-aia-art101-gpai-fines-2"],"attributes":{"provision":{"kind":"decision_authority","key":"aia-gpai-provider-fines","parties":[{"role":"decides","actor":"Commission"}]}}},{"id":"ec-g02-hanson-economic-growth-given-machine-intelligence-b","statement":"Hanson, “Economic Growth Given Machine Intelligence” (2000-08-19) bears on G02. In a variant where technology improves with production experience, some parameter values remove diminishing returns and growth keeps accelerating. The author sets this aside to stay \"conservative\".","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-g02-g"],"attributes":{"thought":{"roles":["context"],"register":"nonfiction","missing":"The feedback runs through learning by doing in production, not through machines doing research.","claimed":{"date":"1998","note":"Author's own citation in the December 2000 paper. The PDF gives the author's Berkeley address; his vita lists a Berkeley postdoc in 1997–1999 and a George Mason post from 1999. The 1998 wording was not retrieved."},"attested":{"date":"2000-08-19","note":"Capture of the PDF (PDF metadata: created 2000-06-22). No earlier capture was retrieved: year-only Wayback requests for this URL and for hanson.berkeley.edu/aigrow.pdf found none, and the CDX index was unavailable during retrieval, so an earlier capture is not excluded."},"tier":{"value":1,"basis":"Tier 1 (unchanged). Working paper with a formal growth model, abstract and references; the checked file gives an econ.berkeley.edu contact address, and the author's December 2000 working paper is issued from the Department of Economics, George Mason University."}}}},{"id":"ec-m07-frey-osborne-2013-future-of-employment-a","statement":"Frey and Osborne, “The Future of Employment: How Susceptible Are Jobs to Computerisation?” (2013-09-17) articulates M07. The authors estimate occupations' susceptibility to computerisation by matching engineering bottleneck variables in O*NET task descriptions with a classifier trained on hand-labelled occupations. The headline estimate: 47 percent of US employment falls in occupations with high probability of computerisation.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"(First passage) Oxford Martin School working paper dated September 17, 2013 (file served by the Oxford Martin School CDN); not the later journal version (Technological Forecasting and Social Change, 2017), whose wording may differ. (Second passage) Oxford Martin School working paper dated September 17, 2013 (file served by the Oxford Martin School CDN); not the later journal version (Technological Forecasting and Social Change, 2017), whose wording may differ. The authors state they 'make no attempt to estimate the number of jobs that will actually be automated'; the figure is technical susceptibility over an unspecified period, not a forecast of job loss. Material quantitative claim: needs targeted recheck.","evidenceIds":["evidence-ec-m07-01","evidence-ec-m07-02"],"attributes":{"thought":{"roles":["earliest-located"],"register":"nonfiction","tier":{"value":1,"basis":"Scholarly working paper with an economist author: Frey, an economic historian at the Oxford Martin School (D2, 2026-10-11, confirmed); Osborne is an engineer."}}}},{"id":"ec-e01-trammell-korinek-2026-growth-under-transformative-ai-nber-revision-a","statement":"Trammell and Korinek, “Economic Growth under Transformative AI” (2023-10) articulates E01. Abstract sentence contrasting automation of production (self-replication) with automation of R&D (self-improvement). Citing the high capital share in AI R&D (Besiroglu et al., 2024), the authors describe a feedback in which automating AI R&D speeds AI capability growth, then cite calibrated models of this feedback (Erdil et al. 2025; Lifland et al. 2025) as having 'speculative elements'. In the risk subsection, the authors contrast a scenario of acceleration from R&D automation alone with scenarios of automated physical production, each with distinct safety concerns.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"(First passage) Wording checked against the April 2026 revision of NBER w31815 (first issued October 2023); whether this wording appears in the October 2023 version was not checked, so it is attested only from April 2026. 'Self-improve' is the authors' gloss on automated R&D in growth models, where the improved object is general technology; the claim is about growth, and the abstract says R&D automation alone may not produce the transformation. (Second passage) Wording checked against the April 2026 revision of NBER w31815 (first issued October 2023); whether this wording appears in the October 2023 version was not checked, so it is attested only from April 2026. The passage cites 2024–2025 work, so this wording is not in the October 2023 version. 'In some sense' hedges the claim; the predictions of near-complete R&D automation are attributed to the cited models, not asserted by the authors. (Third passage) Wording checked against the April 2026 revision of NBER w31815 (first issued October 2023); whether this wording appears in the October 2023 version was not checked, so it is attested only from April 2026. Names a scenario ('software-only singularity') as a risk profile; the term is placed in quotation marks and not developed in the formal models.","evidenceIds":["evidence-ec-e01-28","evidence-ec-e01-29","evidence-ec-e01-31"],"attributes":{"thought":{"roles":["articulation"],"register":"nonfiction","qualifier":"A model scenario under stated conditions, not the author’s own claim.","attested":{"date":"2026-04","note":"Revision date on the cover of the retrieved PDF; the October 2023 wording is recorded under source source-trammell-korinek-2023-growth-under-transformative-ai-nber."},"tier":{"value":1,"basis":"NBER working paper (core venue); Korinek is an economist, Trammell an economist (Oxford, GPI)."},"conditions":"(E01-28) Within the reviewed growth models; acceleration not sufficient 'in isolation'. (E01-29) 'could'; hedged 'in some sense'. (E01-31) Scenario; 'potential for'."}}},{"id":"lg-mx-aldahle-departure-2026","statement":"Ahmad Al-Dahle posted on X on 14 January 2026 that he was joining Airbnb as CTO.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"First-person announcement. The post does not name the Meta role he left; the role attribute uses Meta’s December 2024 byline (context evidence), so it is not established as his title when he left. No end date at Meta is established.","evidenceIds":["lg-ev-mx-aldahle-departure-2026","lg-ev-mx-aldahle-departure-2026-title"],"attributes":{"role":"VP and Head of Generative AI (title from “O futuro da IA: construído com Llama” (Meta (Sala de Imprensa, Brasil)), a December 2024 byline; not named in the post)"}},{"id":"lg-goo-dragan-gdm-safety-alignment","statement":"Anca Dragan's homepage, as retrieved on 15 October 2026, states that she is on leave from UC Berkeley to head AI Safety and Alignment at Google DeepMind.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Self-description on an undated page. A Google post of 2026-10-13 bylines her as VP, AI Safety & Behavior without naming the unit; that title is recorded as context only.","evidenceIds":["lg-ev-goo-dragan-home","lg-ev-goo-dragan-byline-2026"],"attributes":{"role":"head of AI Safety and Alignment"}},{"id":"lg-mx-wu-cofounded-xai","statement":"Business Insider reported on 25 March 2026 that Tony Wu was among xAI's cofounders.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The record rests on reporting; Wu is named in a list of cofounder departures. Not a legal founder record.","evidenceIds":["lg-ev-mx-wu-cofounded-xai"],"attributes":{"role":"Cofounder"}},{"id":"ap-marvin-minsky-minsky-1961-steps","statement":"“Steps Toward Artificial Intelligence” (1961-01) credits Marvin Minsky 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-marvin-minsky-minsky-1961-steps"]},{"id":"ap-eliezer-yudkowsky-lw-2008-what-i-think-if-not-why","statement":"“What I Think, If Not Why” (2008-12-11) 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-what-i-think-if-not-why"]},{"id":"cmp-fsf-2024-d1-persuasion","statement":"No passage meeting the comparison question “Persuasion or manipulation” was found in FSF · 2024 within the inspected scope.","derivation":"inferred","resolution":"supported","inferenceRationale":"Premises: the comparison question asks for the following. The edition names persuasion or manipulation of people as a tracked domain. The inspected scope of FSF · 2024 was: Whole text layer, PDF pp. 1–7 (all sections, both tables, footnotes and acknowledgements read). Method: the edition was read in full and its text layer searched case-insensitively for persua, manipul, decei/decept, influenc, belief. Hits examined and not counted: No hits for any term in the text layer. The four named risk domains (PDF pp. 1, 2, 5) and the Future work list of candidate risks (PDF p. 6) do not include persuasion or manipulation. The conclusion is bounded by that scope and method.","editorialState":"seed","qualification":"A recorded absence within the stated scope: not a claim that the provision was deleted, and not a statement about practice or intent.","evidenceIds":["ev-cmp-fsf-2024-d1-persuasion"],"attributes":{"provision":{"kind":"risk_domain","key":"comparison-d1-persuasion","facet":{"id":"D1.persuasion"}}}},{"id":"cmp-fsf-v2-0-d8-process","statement":"FSF · v2.0 states a provision on the comparison question “Revision process and approval”.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Future Work (PDF p. 8) adds that substantive revisions will be published as appropriate. No approval body or fixed interval for revisions is named. 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-d8-process-1"],"attributes":{"provision":{"kind":"revision","key":"comparison-d8-process","facet":{"id":"D8.process","status":"stated"}}}},{"id":"cmp-faif-2025-d7-internal","statement":"FAIF · December 2025 edition states a provision on the comparison question “Internal reporting”, in a weaker form than the question’s criterion.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Stated, partly: the passage has what the question’s criterion asks for, but in a weaker form than the criterion’s own words. No internal recipient (leadership, board or compliance function) is named for reports or audit results. 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-d7-internal-1","ev-cmp-faif-2025-d7-internal-2"],"attributes":{"provision":{"kind":"reporting","key":"comparison-d7-internal","facet":{"id":"D7.internal","status":"partly"}}}},{"id":"cmp-faif-2026-d8-changes","statement":"No passage meeting the comparison question “Stated changes from the previous edition” was found in FAIF · June 2026 edition within the inspected scope.","derivation":"inferred","resolution":"supported","inferenceRationale":"Premises: the comparison question asks for the following. The edition’s own statement (changelog, introduction or summary) of what changed from its predecessor. Not applicable to the first edition in a sequence. The inspected scope of FAIF · June 2026 edition was: Whole text layer, PDF pp. 1–9 (p. 10 blank): all sections read. Method: the edition was read in full and its text layer searched case-insensitively for previous, December, version, revis, chang, replace. Hits examined and not counted: The edition has no change log or summary of changes and does not mention the December 2025 edition. PDF p. 2 and PDF p. 7, where the edition says xAI may change its approach, describe possible future change; PDF p. 2, on versions of a model and on material change, concerns models; PDF p. 8, on changes to systemic factors, concerns post-mortems of incidents. Confirmed not found on 2026-10-26 by a cross-family archive check of the Internet Archive capture of 2026-07-16 (https://web.archive.org/web/20260716003429/https://media.x.ai/v1/website/xai-frontier-artificial-intelligence-framework-30-june-2026-99c40684.pdf), whose SHA-256 matches the recorded publisher copy. That check notes that its blindness for this cell was compromised, because the edition notes stating that no change log exists had been shown to it. The conclusion is bounded by that scope and method.","editorialState":"seed","qualification":"A recorded absence within the stated scope: not a claim that the provision was deleted, and not a statement about practice or intent.","evidenceIds":["ev-cmp-faif-2026-d8-changes"],"attributes":{"provision":{"kind":"revision","key":"comparison-d8-changes","facet":{"id":"D8.changes"}}}},{"id":"c2-bytedance-seed-coder-licence","statement":"The Seed-Coder-8B-Instruct model card at its revision of 6 June 2025 states that the project is licensed under the MIT License; its revisions up to 6 May 2025 had carried the licence metadata apache-2.0.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Read from the Instruct repository only; the Base and Reasoning repositories carry the mit tag in the Hugging Face listing but their card histories were not compared. The record makes no judgement about the earlier metadata beyond stating it; no Seed statement explaining the change was found.","evidenceIds":["c2-ev-bytedance-seed-coder-licence","c2-ev-bytedance-seed-coder-licence-meta","c2-ev-bytedance-seed-coder-licence-earlier"]},{"id":"c2-governance-tc260-asgf-v3-published","statement":"TC260 released the AI Safety Governance Framework 3.0 on 14 September 2026 at the opening ceremony of the 2026 National Cybersecurity Awareness Week.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The 3.0 preface says it was prepared on the basis of versions 1.0 (2024) and 2.0 (2025) under the guidance of the Cyberspace Administration of China; it does not say that 3.0 replaces or withdraws them, so no supersession is recorded. Not evidence of legal force or implementation.","evidenceIds":["c2-ev-governance-tc260-asgf-v3-published-1","c2-ev-governance-tc260-asgf-v3-published-2"]},{"id":"fl-intl-bletchley-chair-external-evaluations","statement":"The chair's statement on safety testing records that governments recognise their increasing role for seeing that external evaluations are undertaken for frontier AI models developed within their countries, with tests to be conducted before and after deployment.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"States a recognised role 'in accordance with their locally applicable legal frameworks'; it creates no evaluation requirement.","evidenceIds":["evidence-fl-intl-bletchley-chair-external-evaluations-1"],"attributes":{"provision":{"kind":"required_action","key":"intl-bletchley-chair-external-evaluations"}}},{"id":"cpo-dc-aws-gov-50bn","statement":"Amazon announced on 24 November 2025 an investment of up to $50 billion to expand AI and supercomputing capabilities for AWS US government customers, set to break ground in 2026.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"An \"up to\" plan; groundbreaking in 2026 is a target, and the stated addition of nearly 1.3 gigawatts of AI and supercomputing capacity is a planned figure, not built capacity. No site is named.","evidenceIds":["evidence-cpo-dc-aws-gov-50bn-1","evidence-cpo-dc-aws-gov-50bn-2"],"attributes":{"transactionType":"infrastructure_spend","relationshipContext":"external","transactionStage":"planned","amount":{"value":50000000000,"currency":"USD","bound":"up_to","scope":"programme","scopeNote":"as stated"},"statedDates":[{"kind":"construction_start_target","date":"2026","statedBy":"amazon"}]}},{"id":"cpo-dc-stargate-milam-announced","statement":"The September 2025 release announced a Stargate site in Milam County, Texas, to be developed through the SoftBank–OpenAI partnership with SB Energy, a SoftBank Group company.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"An announcement only; no construction start, site capacity or operation date is stated for Milam County.","evidenceIds":["evidence-cpo-dc-stargate-milam-announced-1","evidence-cpo-dc-stargate-milam-announced-2"],"attributes":{"transactionStage":"announced"}},{"id":"ab-ev-usaisi-claude35-tooling-pilot","statement":"The joint pre-deployment evaluation summary for the upgraded Claude 3.5 Sonnet states that US AISI piloted an evaluation method that augments the model with access to bioinformatic tools, and that when tested with Sonnet 3.5 this resulted in increased performance beyond the model alone, at times exceeding measured human expert baselines.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"A piloted method and its stated outcome, for this model and these question sets only. It is not a finding about real-world capability, and the summary gives no figures; the same caveats as the summary's other biological findings apply, including that they rest on US AISI's testing alone. The tools are referred to only as the summary refers to them; no tool, task or workflow is recorded.","evidenceIds":["evidence-ab-ev-usaisi-claude35-tooling-pilot-1"],"attributes":{"evaluatedSystem":"The upgraded Claude 3.5 Sonnet, augmented with access to bioinformatic tools","evaluation":{"what":"Whether augmenting the model with access to bioinformatic tools changes its performance on biology research task questions","how":"A piloted evaluation method giving the model access to such tools, compared with the model alone","result":"Increased performance beyond the model alone, at times exceeding measured human expert baselines","caveats":"Described as a pilot; the summary publishes no figures, states that the findings are from US AISI's testing only and that smaller differences may fall within the margins of error."}}},{"id":"ab-ev-claude-sonnet45-bio-categories","statement":"The Claude Sonnet 4.5 system card states that its CBRN evaluations primarily focus on biological risks with the largest consequences and sets out separate ASL-3 and ASL-4 threat models for that domain.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Records the evaluation category and the stated focus, not a finding about the model. The thresholds are recorded at the generality of the card's own headings; the dataset already holds the corresponding RSP provisions.","evidenceIds":["evidence-ab-ev-claude-sonnet45-bio-categories-1","evidence-ab-ev-claude-sonnet45-bio-categories-2"]},{"id":"ab-ev-anthropic-hub-sonnet55-cb1","statement":"Anthropic's transparency hub, as served on 28 October 2026, states that it deployed Claude Sonnet 5.5 with the same chemical and biological misuse protections as Claude Opus 5, treating it as meeting its CB-1 threshold for models able to significantly help someone with a basic technical background make known chemical or biological weapons.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"A summary page, not the system card itself, and a mutable one: the wording may change without notice, and the record is bounded to the snapshot read. It is the developer's own threshold treatment, not an independent finding, and not evidence that the protections work. The threshold is recorded at the generality of the page's own wording.","evidenceIds":["evidence-ab-ev-anthropic-hub-sonnet55-cb1-1"],"attributes":{"evaluatedSystem":"Claude Sonnet 5.5","evaluation":{"what":"Whether the model meets Anthropic's CB-1 chemical and biological threshold","how":"The page refers to Anthropic's own assessments and to the summaries of key safety evaluations in the model's system card","result":"Treated as meeting the CB-1 threshold and deployed with the same chemical and biological misuse protections as Claude Opus 5","caveats":"The page states it gives summaries only and refers readers to the full system card for the complete reported results; the system card itself was not retrieved for this record."}}}],"sources":[{"id":"source-fsf-1-0","title":"Frontier Safety Framework — v1.0","authors":["Google DeepMind"],"publisher":"Google DeepMind","publicationDate":"2024-05-17","version":"Frontier Safety Framework v1.0; publication date 2024-05-17.","versionDate":"2024-05-17","canonicalUrl":"https://storage.googleapis.com/deepmind-media/DeepMind.com/Blog/introducing-the-frontier-safety-framework/fsf-technical-report.pdf","evidenceUrl":"https://storage.googleapis.com/deepmind-media/DeepMind.com/Blog/introducing-the-frontier-safety-framework/fsf-technical-report.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20240517140902/https://storage.googleapis.com/deepmind-media/DeepMind.com/Blog/introducing-the-frontier-safety-framework/fsf-technical-report.pdf","archivedAt":"2024-05-17","accessedAt":"2026-09-18","reuse":{"attribution":"Google DeepMind, Frontier Safety Framework — v1.0.","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-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-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-faif-2026","title":"Frontier Artificial Intelligence Framework — June 2026 edition","authors":["xAI"],"publisher":"xAI","publicationDate":null,"version":"Frontier Artificial Intelligence Framework June 2026 edition; effective date 2026-06-30. Visible public cover identifies the effective edition; stale embedded PDF draft metadata is not used to infer publication status.","versionDate":"2026-06-30","canonicalUrl":"https://media.x.ai/v1/website/xai-frontier-artificial-intelligence-framework-30-june-2026-99c40684.pdf","evidenceUrl":"https://media.x.ai/v1/website/xai-frontier-artificial-intelligence-framework-30-june-2026-99c40684.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20260716003429/https://media.x.ai/v1/website/xai-frontier-artificial-intelligence-framework-30-june-2026-99c40684.pdf","archivedAt":"2026-07-16","accessedAt":"2026-09-18","reuse":{"attribution":"xAI, Frontier Artificial Intelligence Framework — June 2026 edition.","licence":"No open redistribution licence asserted","note":"Selekoa practice: metadata, attributed paraphrases and short attributed quotations under the editorial policy (at most 50 words each; at most 400 quoted words per document across the dataset, and about 10% of a document under about 2,000 words). Documents are linked externally, not redistributed. Publisher terms that forbid quotation override this practice."}},{"id":"source-amazon-q3-2025","title":"Amazon.com announces third quarter results","authors":["Amazon"],"publisher":"Amazon","publicationDate":"2025-10-30","version":"Dated announcement inspected 2026-09-19; a historical company statement, not independent verification of implementation.","canonicalUrl":"https://www.aboutamazon.com/news/company-news/amazon-earnings-q3-2025-report","evidenceUrl":"https://www.aboutamazon.com/news/company-news/amazon-earnings-q3-2025-report","language":"en","archivedUrl":"https://web.archive.org/web/20260515205133/https://www.aboutamazon.com/news/company-news/amazon-earnings-q3-2025-report","archivedAt":"2026-05-15","accessedAt":"2026-09-19","reuse":{"attribution":"Amazon, Amazon.com announces third quarter results (2025-10-30).","licence":"No open redistribution licence asserted","note":"Bibliographic metadata and editorial paraphrases only; no full source redistributed."}},{"id":"source-cn-keg-search","title":"计算机系知识工程实验室联合搜狗与智谱AI发布全新学术搜索产品","authors":["Tsinghua University Department of Computer Science"],"publisher":"Tsinghua University Department of Computer Science","publicationDate":"2021-06-03","version":"Source language: Simplified Chinese (zh-Hans). Dated university account of the 2021-06-02 event.","canonicalUrl":"https://www.cs.tsinghua.edu.cn/info/1088/4451.htm","evidenceUrl":"https://www.cs.tsinghua.edu.cn/info/1088/4451.htm","language":"zh-Hans","archivedUrl":"https://web.archive.org/web/20260615011410/https://www.cs.tsinghua.edu.cn/info/1088/4451.htm","archivedAt":"2026-06-15","accessedAt":"2026-09-29","reuse":{"attribution":"Tsinghua University Department of Computer Science, 计算机系知识工程实验室联合搜狗与智谱AI发布全新学术搜索产品 (2021-06-03).","licence":"No redistribution licence asserted for this source expression","note":"Bibliographic metadata and brief attributed paraphrases only; model licensing is distinct from source-document reuse."}},{"id":"source-minsky-1961-mit","title":"Steps Toward Artificial Intelligence (MIT-hosted transcription)","authors":["Marvin Minsky"],"publisher":"Proceedings of the IRE","publicationDate":"1961-01","version":"Later HTML transcription; introduction and summary of the argument read by the pre-1996 candidate search. Transcription hosted at web.mit.edu.","canonicalUrl":"https://web.mit.edu/dxh/www/marvin/web.media.mit.edu/~minsky/papers/steps.html","evidenceUrl":"https://web.mit.edu/dxh/www/marvin/web.media.mit.edu/~minsky/papers/steps.html","language":"en","archivedUrl":"https://web.archive.org/web/20261005100457/https://web.mit.edu/dxh/www/marvin/web.media.mit.edu/~minsky/papers/steps.html","archivedAt":"2026-10-05","accessedAt":"2026-10-06","reuse":{"attribution":"Marvin Minsky, \"Steps Toward Artificial Intelligence\", Proceedings of the IRE 49 (1961).","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-eoc-1997-references","title":"Engines of Creation, notes and references (web edition)","authors":["K. Eric Drexler"],"publisher":"Foresight Institute (web edition of Engines of Creation)","publicationDate":"1986","version":"Web edition page as captured by the Internet Archive on 1997-07-09 (timestamp 19970709104746); it carries the notice \"Copyright 1986, K. Eric Drexler\". Retrieved 2026-10-06.","canonicalUrl":"http://www.foresight.org/EOC/EOC_References.html","evidenceUrl":"https://web.archive.org/web/19970709104746/http://www.foresight.org:80/EOC/EOC_References.html","language":"en","archivedUrl":"https://web.archive.org/web/19970709104746/http://www.foresight.org:80/EOC/EOC_References.html","archivedAt":"1997-07-09","accessedAt":"2026-10-06","reuse":{"attribution":"K. Eric Drexler, Engines of Creation: The Coming Era of Nanotechnology (1986), Foresight Institute web edition.","licence":"All rights reserved, according to the page’s copyright notice.","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"source-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-christiano-2022-where-i-agree-and-disagree","title":"Where I agree and disagree with Eliezer","authors":["Paul Christiano"],"publisher":"Alignment Forum and LessWrong","publicationDate":"2022-06-19","version":"Text checked in the post’s first stored version through the LessWrong GraphQL API on 2026-10-06.","canonicalUrl":"https://www.lesswrong.com/posts/CoZhXrhpQxpy9xw9y/where-i-agree-and-disagree-with-eliezer","evidenceUrl":"https://www.lesswrong.com/posts/CoZhXrhpQxpy9xw9y/where-i-agree-and-disagree-with-eliezer","language":"en","archivedUrl":"https://web.archive.org/web/20220619200048/https://www.lesswrong.com/posts/CoZhXrhpQxpy9xw9y/where-i-agree-and-disagree-with-eliezer","archivedAt":"2022-06-19","accessedAt":"2026-10-06","reuse":{"attribution":"Paul Christiano, \"Where I agree and disagree with Eliezer\", Alignment Forum and LessWrong, 2022-06-19.","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-time-2023-shut-it-all-down","title":"Pausing AI Developments Isn't Enough. We Need to Shut it All Down","authors":["Eliezer Yudkowsky"],"publisher":"TIME","publicationDate":"2023-03-29","version":"TIME, retrieved 2026-10-06 (SHA-256 4a8c77e15456ca68b530df8bd825296e3923f621b0c0b09dacce312ae94182d6). Page metadata datePublished 2023-03-29T22:01:58Z. The page title currently reads \"The Only Way to Deal With the Threat From AI? Shut It Down\"; the headline is the one crossposted to LessWrong on 2023-04-08.","canonicalUrl":"https://time.com/6266923/ai-eliezer-yudkowsky-open-letter-not-enough/","evidenceUrl":"https://time.com/6266923/ai-eliezer-yudkowsky-open-letter-not-enough/","language":"en","archivedUrl":"https://web.archive.org/web/20261003171949/https://time.com/6266923/ai-eliezer-yudkowsky-open-letter-not-enough/","archivedAt":"2026-10-03","accessedAt":"2026-10-06","reuse":{"attribution":"Eliezer Yudkowsky, “Pausing AI Developments Isn't Enough. We Need to Shut it All Down”, TIME, 2023-03-29.","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":"source-extropians-1999-08-04-yudkowsky-1587","title":"Re: Does a parental model work for having A.I. that can be trusted? (Extropians mailing list)","authors":["Eliezer Yudkowsky"],"publisher":"Extropians mailing list","publicationDate":"1999-08-04","version":"Archive HTML as served, retrieved 2026-10-07 (SHA-256 3b2d6b2b2285f03c56382a096fc6ebdaf54003e6bac0124bf96ba0bf3665c7d2); 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.3Q99/1587.html","evidenceUrl":"http://extropians.weidai.com/extropians.3Q99/1587.html","language":"en","archivedUrl":"https://web.archive.org/web/20240912214733/http://extropians.weidai.com/extropians.3Q99/1587.html","archivedAt":"2024-09-12","accessedAt":"2026-10-07","reuse":{"attribution":"Eliezer Yudkowsky, “Re: Does a parental model work for having A.I. that can be trusted?”, Extropians mailing list, 1999-08-04.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"source-lw-2007-thou-art-godshatter","title":"Thou Art Godshatter","authors":["Eliezer Yudkowsky"],"publisher":"Overcoming Bias","publicationDate":"2007-11-13","version":"Post body retrieved through the LessWrong GraphQL API on 2026-10-06 (SHA-256 b6abf133e881a827dca01e79a7c5493835c6d026c00fcdd6d46fd6b08086551d), content version 1.0.0. As imported to LessWrong.","canonicalUrl":"https://www.lesswrong.com/posts/cSXZpvqpa9vbGGLtG/thou-art-godshatter","evidenceUrl":"https://www.lesswrong.com/posts/cSXZpvqpa9vbGGLtG/thou-art-godshatter","language":"en","archivedUrl":"https://web.archive.org/web/20260923090253/https://www.lesswrong.com/posts/cSXZpvqpa9vbGGLtG/thou-art-godshatter","archivedAt":"2026-09-23","accessedAt":"2026-10-06","reuse":{"attribution":"Eliezer Yudkowsky, “Thou Art Godshatter”, Overcoming Bias, as imported to LessWrong, 2007-11-13.","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-lw-2008-what-i-think-if-not-why","title":"What I Think, If Not Why","authors":["Eliezer Yudkowsky"],"publisher":"Overcoming Bias","publicationDate":"2008-12-11","version":"Post body retrieved through the LessWrong GraphQL API on 2026-10-06 (SHA-256 d88333aee9abdfc2c2dc9f6a5ac5401b3a8719b7df75934e2eec9765fe2c471d), content version 1.0.0. As imported to LessWrong.","canonicalUrl":"https://www.lesswrong.com/posts/z3kYdw54htktqt9Jb/what-i-think-if-not-why","evidenceUrl":"https://www.lesswrong.com/posts/z3kYdw54htktqt9Jb/what-i-think-if-not-why","language":"en","archivedUrl":"https://web.archive.org/web/20260927022025/https://www.lesswrong.com/posts/z3kYdw54htktqt9Jb/what-i-think-if-not-why","archivedAt":"2026-09-27","accessedAt":"2026-10-06","reuse":{"attribution":"Eliezer Yudkowsky, “What I Think, If Not Why”, Overcoming Bias, as imported to LessWrong, 2008-12-11.","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-hanson-economic-growth-given-machine-intelligence","title":"Economic Growth Given Machine Intelligence","authors":["Robin Hanson"],"publisher":"Working paper","publicationDate":null,"version":"Wayback capture of http://hanson.gmu.edu/aigrow.pdf (PDF created 2000-06-22). Capture retrieved 2026-10-08 (SHA-256 754380e21b0e9f5a62fe71ea238a86e8640df48a0551c55cf4877a2a732b99c1). Versions: claimed 1998 version (1998 (author's citation)): not located; hanson.gmu.edu/aigrow.pdf (undated; dated by the 2000-08-19 capture): checked. Published version: Hanson, R. (n.d.). Economic growth given machine intelligence [Unpublished manuscript]. Retrieved from a 2000-08-19 Wayback capture of hanson.gmu.edu/aigrow.pdf. No published version located (Crossref search 2026-10-10 found none).","canonicalUrl":"https://mason.gmu.edu/~rhanson/aigrow.pdf","evidenceUrl":"https://web.archive.org/web/20000819034042/http://hanson.gmu.edu/aigrow.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20000819034042/http://hanson.gmu.edu/aigrow.pdf","archivedAt":"2000-08-19","accessedAt":"2026-10-08","reuse":{"attribution":"Robin Hanson, “Economic Growth Given Machine Intelligence”, Working paper, author's site (hanson.gmu.edu/aigrow.pdf).","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"source-frey-osborne-2013-future-of-employment","title":"The Future of Employment: How Susceptible Are Jobs to Computerisation?","authors":["Carl Benedikt Frey","Michael A. Osborne"],"publisher":"Oxford Martin School working paper","publicationDate":"2013-09-17","version":"Full text (current file served by the Oxford Martin School; requested at oxfordmartin.ox.ac.uk, which redirected to this host). Current copy retrieved directly on 2026-10-09; no archive capture (Wayback unreachable), so earlier bytes are not authenticated. SHA-256 1d0ef561ebef9ececda63403a40860aa172be953fe2b050d27de8987c9e852fc. Journal version not retrieved. Frey, C. B., & Osborne, M. A. (2017). The future of employment: How susceptible are jobs to computerisation? Technological Forecasting and Social Change, 114, 254–280. https://doi.org/10.1016/j.techfore.2016.08.019","canonicalUrl":"https://oms-www.files.svdcdn.com/production/downloads/academic/The_Future_of_Employment.pdf","evidenceUrl":"https://oms-www.files.svdcdn.com/production/downloads/academic/The_Future_of_Employment.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20261008065923/https://oms-www.files.svdcdn.com/production/downloads/academic/The_Future_of_Employment.pdf","archivedAt":"2026-10-08","accessedAt":"2026-10-09","reuse":{"attribution":"Carl Benedikt Frey, Michael A. Osborne, “The Future of Employment: How Susceptible Are Jobs to Computerisation?”, Oxford Martin School working paper, 2013-09-17.","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-trammell-korinek-2026-growth-under-transformative-ai-nber-revision","title":"Economic Growth under Transformative AI","authors":["Philip Trammell","Anton Korinek"],"publisher":"NBER Working Paper 31815","publicationDate":"2023-10","version":"Current NBER PDF, the April 2026 revision. The October 2023 version was retrieved separately from a Wayback capture of 2023-11-29 (source source-trammell-korinek-2023-growth-under-transformative-ai-nber); their wording differs. SHA-256 06dda75180302720e0843a180c00d04fa99e38db0f5d6d4e06d7f29388a115bb. Version a reader would cite, or full text, not retrieved. Trammell, P., & Korinek, A. (2023). Economic growth under transformative AI (NBER Working Paper No. 31815, original October 2023 version). https://doi.org/10.3386/w31815","canonicalUrl":"https://www.nber.org/system/files/working_papers/w31815/w31815.pdf","evidenceUrl":"https://www.nber.org/system/files/working_papers/w31815/w31815.pdf","language":"en","accessedAt":"2026-10-11","reuse":{"attribution":"Philip Trammell, Anton Korinek, “Economic Growth under Transformative AI”, NBER Working Paper 31815, 2023-10.","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":"lg-src-goo-dragan-home","title":"Anca Dragan (personal academic homepage, undated)","authors":["Anca Dragan"],"publisher":"UC Berkeley EECS","publicationDate":null,"version":"Undated post, retrieved on 2026-10-15 (SHA-256 cbbae006372c2d5abd6b937251f9d83398c105cc5c3252c36bb6acb7e2afaa76) directly from the publisher. Its wording can change; earlier wording is not authenticated.","canonicalUrl":"https://people.eecs.berkeley.edu/~anca/","evidenceUrl":"https://people.eecs.berkeley.edu/~anca/","language":"en","archivedUrl":"https://web.archive.org/web/20260925150734/https://people.eecs.berkeley.edu/~anca/","archivedAt":"2026-09-25","accessedAt":"2026-10-15","reuse":{"attribution":"Anca Dragan, “Anca Dragan (personal academic 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-goo-gemini4-safety-2026","title":"Developing Gemini 4 Argon responsibly and safely","authors":["Anca Dragan","Raluca Ada Popa"],"publisher":"Google","publicationDate":"2026-10-13","version":"Dated page, retrieved on 2026-10-15 (SHA-256 b588bfd8e3d2d7207e383c354e39802c85a4d68af1c354fb60f38f2a242d6854) directly from the publisher.","canonicalUrl":"https://blog.google/innovation-and-ai/technology/safety-security/gemini-4-model-safety-security/","evidenceUrl":"https://blog.google/innovation-and-ai/technology/safety-security/gemini-4-model-safety-security/","language":"en","archivedUrl":"https://web.archive.org/web/20261015100606/https://blog.google/innovation-and-ai/technology/safety-security/gemini-4-model-safety-security/","archivedAt":"2026-10-15","accessedAt":"2026-10-15","reuse":{"attribution":"Anca Dragan, Raluca Ada Popa, “Developing Gemini 4 Argon responsibly and safely”, 2026-10-13.","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-mx-aldahle-departure","title":"Post of 14 January 2026 on joining Airbnb (X)","authors":["Ahmad Al-Dahle"],"publisher":"X","publicationDate":"2026-01-14","version":"Dated post, retrieved on 2026-10-15 (SHA-256 e72dda9a9eb14ce0faf2abeb642bdf77d24b06004eeecabc96f1bb1d10cc8590) through X’s own syndication endpoint for the post’s text, which truncates long notes.","canonicalUrl":"https://x.com/ahmad_al_dahle/status/2011440460821320056","evidenceUrl":"https://x.com/ahmad_al_dahle/status/2011440460821320056","language":"en","archivedUrl":"https://web.archive.org/web/20260114140911/https://twitter.com/Ahmad_Al_Dahle/status/2011440460821320056","archivedAt":"2026-01-14","accessedAt":"2026-10-15","reuse":{"attribution":"Ahmad Al-Dahle, “Post of 14 January 2026 on joining Airbnb (X)”, 2026-01-14.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"lg-src-mx-bi-kroiss","title":"Another Co-Founder Is Exiting Elon Musk's XAI","authors":["Grace Kay"],"publisher":"Business Insider","publicationDate":"2026-03-25","version":"Dated page, retrieved on 2026-10-15 (SHA-256 6ca85271f8ae9a067db52f8ffdfdf07f859ee1e1e2238d06fa29e878997eaa0f) directly from the publisher.","canonicalUrl":"https://www.businessinsider.com/manuel-kroiss-xai-cofounder-departure-elon-musk-2026-3","evidenceUrl":"https://www.businessinsider.com/manuel-kroiss-xai-cofounder-departure-elon-musk-2026-3","language":"en","archivedUrl":"https://web.archive.org/web/20260808024547/https://www.businessinsider.com/manuel-kroiss-xai-cofounder-departure-elon-musk-2026-3","archivedAt":"2026-08-08","accessedAt":"2026-10-15","reuse":{"attribution":"Grace Kay, “Another Co-Founder Is Exiting Elon Musk's XAI”, 2026-03-25.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"lg-src-mx-meta-br-llama-2024","title":"O futuro da IA: construído com Llama","authors":["Ahmad Al-Dahle"],"publisher":"Meta","publicationDate":"2024-12-19","version":"Dated page, retrieved on 2026-10-15 (SHA-256 444da7cda6709167042f76fb5f9987834df1e8bd9e2aba2d7ae2af9527a2466f) directly from the publisher. Publisher as printed on the inspected expression: Meta (Sala de Imprensa, Brasil).","canonicalUrl":"https://about.fb.com/br/news/2024/12/o-futuro-da-ia-construido-com-llama/","evidenceUrl":"https://about.fb.com/br/news/2024/12/o-futuro-da-ia-construido-com-llama/","language":"pt","archivedUrl":"https://web.archive.org/web/20260614192353/https://about.fb.com/br/news/2024/12/o-futuro-da-ia-construido-com-llama/","archivedAt":"2026-06-14","accessedAt":"2026-10-15","reuse":{"attribution":"Ahmad Al-Dahle, “O futuro da IA: construído com Llama”, 2024-12-19.","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":"c2-src-seed-coder-hf-readme-1c5af72a","title":"Seed-Coder-8B-Instruct model card (README.md, revision 1c5af72a)","authors":["ByteDance Seed"],"publisher":"ByteDance Seed (Hugging Face organisation ByteDance-Seed)","publicationDate":"2025-06-06","version":"Dated page, retrieved on 2026-10-27 (SHA-256 e6dafc62c9964279ba786fba00e0a49421a072d4e4bdee2a8d74bc7215647353) directly from the publisher. Version: 1c5af72afa21c22c9272a0b60a3a404195e88fcd. README.md at the Hugging Face commit of 6 June 2025 (latest on main at retrieval).","canonicalUrl":"https://huggingface.co/ByteDance-Seed/Seed-Coder-8B-Instruct/raw/1c5af72afa21c22c9272a0b60a3a404195e88fcd/README.md","evidenceUrl":"https://huggingface.co/ByteDance-Seed/Seed-Coder-8B-Instruct/raw/1c5af72afa21c22c9272a0b60a3a404195e88fcd/README.md","language":"en","archivedUrl":"https://web.archive.org/web/20261028224015/https://huggingface.co/ByteDance-Seed/Seed-Coder-8B-Instruct/raw/1c5af72afa21c22c9272a0b60a3a404195e88fcd/README.md","archivedAt":"2026-10-28","accessedAt":"2026-10-27","reuse":{"attribution":"ByteDance Seed, “Seed-Coder-8B-Instruct model card (README.md, revision 1c5af72a)”, 2025-06-06.","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":"c2-src-seed-coder-hf-readme-f80ca53f","title":"Seed-Coder-8B-Instruct model card (README.md, revision f80ca53f)","authors":["ByteDance Seed"],"publisher":"ByteDance Seed (Hugging Face organisation ByteDance-Seed)","publicationDate":"2025-04-30","version":"Dated page, retrieved on 2026-10-27 (SHA-256 ed3f4cd98b7933cf50c318b22b9d3fc1c49281fa3a5d137a3e768bd07fe2bf55) directly from the publisher. Version: f80ca53f4789f6eb8e73e1bf633291d875dd56d1. Earlier README.md revision of 30 April 2025; whether the repository was public at that date is not established.","canonicalUrl":"https://huggingface.co/ByteDance-Seed/Seed-Coder-8B-Instruct/raw/f80ca53f4789f6eb8e73e1bf633291d875dd56d1/README.md","evidenceUrl":"https://huggingface.co/ByteDance-Seed/Seed-Coder-8B-Instruct/raw/f80ca53f4789f6eb8e73e1bf633291d875dd56d1/README.md","language":"en","archivedUrl":"https://web.archive.org/web/20261028221627/https://huggingface.co/ByteDance-Seed/Seed-Coder-8B-Instruct/raw/f80ca53f4789f6eb8e73e1bf633291d875dd56d1/README.md","archivedAt":"2026-10-28","accessedAt":"2026-10-27","reuse":{"attribution":"ByteDance Seed, “Seed-Coder-8B-Instruct model card (README.md, revision f80ca53f)”, 2025-04-30.","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":"c2-src-tc260-asgf-v3-news","title":"《人工智能安全治理框架3.0》发布","authors":["全国网络安全标准化技术委员会"],"publisher":"全国网络安全标准化技术委员会","publicationDate":"2026-09-14","version":"Dated page, retrieved on 2026-10-27 (SHA-256 14b77957e1033558bd278431c1ed07f3525cf8af9ba77a89c04a443a31552a65) directly from the publisher. TC260 news item dated 2026-09-14, read as it stood on 2026-10-27. Its wording can change; earlier wording is not authenticated. English title (machine translation): “AI Safety Governance Framework 3.0 released”. Publisher as printed on the inspected expression: 全国网络安全标准化技术委员会 (tc260.org.cn).","canonicalUrl":"https://www.tc260.org.cn/tc260/xwdt1/202609/d513a007d04347f58e483fabaefb34b8.shtml","evidenceUrl":"https://www.tc260.org.cn/tc260/xwdt1/202609/d513a007d04347f58e483fabaefb34b8.shtml","language":"zh-Hans","archivedUrl":"https://web.archive.org/web/20260923081457/https://www.tc260.org.cn/tc260/xwdt1/202609/d513a007d04347f58e483fabaefb34b8.shtml","archivedAt":"2026-09-23","accessedAt":"2026-10-27","reuse":{"attribution":"全国网络安全标准化技术委员会, “《人工智能安全治理框架3.0》发布”, 2026-09-14.","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":"c2-src-tc260-asgf-v3-pdf","title":"人工智能安全治理框架3.0","authors":["全国网络安全标准化技术委员会"],"publisher":"全国网络安全标准化技术委员会","publicationDate":"2026-09","version":"Dated document, retrieved on 2026-10-27 (SHA-256 cff709d8b1b2a0dfd2ff24df4d6b942b160975331e44c918e14a47d5c86da43d) directly from the publisher. Version: 《人工智能安全治理框架3.0》.pdf.pdf. Bilingual PDF attached to TC260's news item of 14 September 2026 (Chinese text followed by the publisher's English text). Quotes are from the Chinese text. English title (machine translation): “AI Safety Governance Framework 3.0”. Publisher as printed on the inspected expression: 全国网络安全标准化技术委员会 (tc260.org.cn).","canonicalUrl":"https://www.tc260.org.cn/tc260/xwdt1/202609/d513a007d04347f58e483fabaefb34b8/files/%E3%80%8A%E4%BA%BA%E5%B7%A5%E6%99%BA%E8%83%BD%E5%AE%89%E5%85%A8%E6%B2%BB%E7%90%86%E6%A1%86%E6%9E%B63.0%E3%80%8B.pdf.pdf","evidenceUrl":"https://www.tc260.org.cn/tc260/xwdt1/202609/d513a007d04347f58e483fabaefb34b8/files/%E3%80%8A%E4%BA%BA%E5%B7%A5%E6%99%BA%E8%83%BD%E5%AE%89%E5%85%A8%E6%B2%BB%E7%90%86%E6%A1%86%E6%9E%B63.0%E3%80%8B.pdf.pdf","language":"zh-Hans","archivedUrl":"https://web.archive.org/web/20260920144343/https://www.tc260.org.cn/tc260/xwdt1/202609/d513a007d04347f58e483fabaefb34b8/files/%E3%80%8A%E4%BA%BA%E5%B7%A5%E6%99%BA%E8%83%BD%E5%AE%89%E5%85%A8%E6%B2%BB%E7%90%86%E6%A1%86%E6%9E%B63.0%E3%80%8B.pdf.pdf","archivedAt":"2026-09-20","accessedAt":"2026-10-27","reuse":{"attribution":"全国网络安全标准化技术委员会, “人工智能安全治理框架3.0”, 2026-09.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"source-bletchley-chair-safety-testing-2023","title":"Safety Testing: Chair's Statement of Session Outcomes, 2 November 2023","authors":["Chair of the AI Safety Summit (UK government)"],"publisher":"Department for Science, Innovation and Technology","publicationDate":"2023-11-02","version":"Retrieved on 2026-10-27 (SHA-256 c7b3a9cb614c69a77061b04735d88697021a21ad583ba738a5b11befe7d93e38). GOV.UK HTML policy paper, 'Published 2 November 2023', as served on 2026-10-27. 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-chairs-statement-safety-testing-2-november/safety-testing-chairs-statement-of-session-outcomes-2-november-2023","evidenceUrl":"https://www.gov.uk/government/publications/ai-safety-summit-2023-chairs-statement-safety-testing-2-november/safety-testing-chairs-statement-of-session-outcomes-2-november-2023","language":"en","accessedAt":"2026-10-27","reuse":{"attribution":"Chair of the AI Safety Summit (UK government), “Safety Testing: Chair's Statement of Session Outcomes, 2 November 2023”, 2023-11-02.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"source-nist-preeval-claude-35-sonnet-2024","title":"Pre-Deployment Evaluation of Anthropic’s Upgraded Claude 3.5 Sonnet (November 19, 2024)","authors":["National Institute of Standards and Technology (U.S. AI Safety Institute)"],"publisher":"National Institute of Standards and Technology","publicationDate":"2024-11-19","version":"Retrieved on 2026-10-27 (SHA-256 1b9a64eb9d1953bdd916767279b7021cabc79fc837018b2f75fc3ccf2c8f983a). The cross-family check of 2026-10-28 opened the Internet Archive capture linked here and matched the quoted wording in it, the release-period expression (decision 9); a capture shows only what the archive retrieved that day.","canonicalUrl":"https://www.nist.gov/news-events/news/2024/11/pre-deployment-evaluation-anthropics-upgraded-claude-35-sonnet","evidenceUrl":"https://www.nist.gov/news-events/news/2024/11/pre-deployment-evaluation-anthropics-upgraded-claude-35-sonnet","language":"en","archivedUrl":"https://web.archive.org/web/20241119153906/https://www.nist.gov/news-events/news/2024/11/pre-deployment-evaluation-anthropics-upgraded-claude-35-sonnet","archivedAt":"2024-11-19","accessedAt":"2026-10-27","reuse":{"attribution":"National Institute of Standards and Technology (U.S. AI Safety Institute), “Pre-Deployment Evaluation of Anthropic’s Upgraded Claude 3.5 Sonnet (November 19, 2024)”, 2024-11-19.","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-dc-sb-stargate-five-sites-2025","title":"OpenAI, Oracle, and SoftBank expand Stargate with five new AI data center sites","authors":["SoftBank Group Corp. (joint release of OpenAI, Oracle and SoftBank Group)"],"publisher":"SoftBank Group Corp. (joint release of OpenAI, Oracle and SoftBank Group)","publicationDate":"2025-09-23","version":"Retrieved for the candidate pack on 2026-10-28 (SHA-256 1a993c3440862bb1f2157560f56f5df305cbf48be9346370ebd9be533d635d93). Press release page as served on 2026-10-28; dated \"September 24, 2025 (Texas, September 23, 2025)\" and attributed to OpenAI, Oracle and SoftBank Group Corp. OpenAI's own post (openai.com/index/five-new-stargate-sites/) refused retrieval; the cross-family checker should open it or its capture and compare. Retrieved again at admission on 2026-10-30 (SHA-256 a3d77f8db28800481eac9625ed0caf759b0f12015cf3b6f9f139cd330cc28292). Its wording can change; earlier wording is not authenticated.","canonicalUrl":"https://group.softbank/en/news/press/20250924","evidenceUrl":"https://group.softbank/en/news/press/20250924","language":"en","accessedAt":"2026-10-28","reuse":{"attribution":"SoftBank Group Corp. (joint release of OpenAI, Oracle and SoftBank Group), “OpenAI, Oracle, and SoftBank expand Stargate with five new AI data center sites”, 2025-09-23.","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-cpo-dc-amazon-gov-50bn-2025","title":"Amazon to invest up to $50 billion to expand AI and supercomputing infrastructure for US government agencies","authors":["Amazon (About Amazon)"],"publisher":"Amazon (About Amazon)","publicationDate":"2025-11-24","version":"Retrieved for the candidate pack on 2026-10-28 (SHA-256 33cbc1c32a11cb15fd4a90e6dd21b6b1dbc23bedb94d78e37e7cb379131211fe). News page as served on 2026-10-28. The visible text prints no date; the page metadata gives datePublished 2025-11-24T16:00:09Z. Retrieved again at admission on 2026-10-30 (SHA-256 c2c224718427c46f611b0503ff9c9a11ac3c1bcc5a202d45b6efd301e74f6e3a). Its wording can change; earlier wording is not authenticated.","canonicalUrl":"https://www.aboutamazon.com/news/company-news/amazon-ai-investment-us-federal-agencies","evidenceUrl":"https://www.aboutamazon.com/news/company-news/amazon-ai-investment-us-federal-agencies","language":"en","accessedAt":"2026-10-28","reuse":{"attribution":"Amazon (About Amazon), “Amazon to invest up to $50 billion to expand AI and supercomputing infrastructure for US government agencies”, 2025-11-24.","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-ab-claude-sonnet45-card","title":"System Card: Claude Sonnet 4.5","authors":["Anthropic"],"publisher":"Anthropic","publicationDate":null,"version":"Retrieved for the candidate pack on 2026-10-28 (SHA-256 d5d1ce28fea101ae3b201a5d4da100090a5d6b73246f7da820cafd7dbb11100e). PDF whose cover reads 'System Card: Claude Sonnet 4.5, September 2025', as served on 2026-10-28; the request redirected to https://www-cdn.anthropic.com/963373e433e489a87a10c823c52a0a013e9172dd.pdf. No day of the month is stated. Retrieved again at admission on 2026-10-30 (SHA-256 d5d1ce28fea101ae3b201a5d4da100090a5d6b73246f7da820cafd7dbb11100e, the same bytes).","canonicalUrl":"https://assets.anthropic.com/m/12f214efcc2f457a/original/Claude-Sonnet-4-5-System-Card.pdf","evidenceUrl":"https://assets.anthropic.com/m/12f214efcc2f457a/original/Claude-Sonnet-4-5-System-Card.pdf","language":"en","accessedAt":"2026-10-28","reuse":{"attribution":"Anthropic, “System Card: Claude Sonnet 4.5”.","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-anthropic-hub","title":"Transparency hub (Anthropic), model-by-model safety summaries","authors":["Anthropic"],"publisher":"Anthropic","publicationDate":null,"version":"Retrieved for the candidate pack on 2026-10-28 (SHA-256 ddbdde44443a71db8c87d1d710edb263b5c85583ae4917a9858e5bdbcaaa266e). The page as served on 2026-10-28, a mutable hub that summarises the current model line (Claude Mythos 5.1 down to Claude Sonnet 3.7). It states no publication date for the page itself. Retrieved again at admission on 2026-10-30 (SHA-256 180bb54e99e373f383f2d06ea3975f8819bd2708bc49a9f5fafd088074259a63). Its wording can change; earlier wording is not authenticated.","canonicalUrl":"https://www.anthropic.com/transparency","evidenceUrl":"https://www.anthropic.com/transparency","language":"en","accessedAt":"2026-10-28","reuse":{"attribution":"Anthropic, “Transparency hub (Anthropic), model-by-model safety summaries”.","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-anthropic-rainier-2025","sourceRole":"primary","support":"supports","locator":"Highlights since the previous earnings announcement, Project Rainier bullet","passage":"Launched Project Rainier, a massive AI compute cluster containing nearly 500,000 Trainium2 chips, to build and deploy Anthropic’s leading Claude AI models.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-15","note":"Re-checked 2026-10-15 against the retrieved page (research-debt pass resolving Audit B's RESEARCH MORE); quotation verified verbatim by a machine agent. Company statements are not independent proof of implementation. Human source review pending."},{"id":"evidence-cn-keg-tsinghua","sourceRole":"primary","support":"supports","locator":"Title and paragraph quoting 李涓子 beginning 知识工程实验室","passage":"The department describes KEG as a long-standing computer-science research team.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-09-29","note":"English paraphrase translated by machine from the cited original Chinese passage; no human translation review. Exploratory seed, not editorially accepted. Translation scope and ambiguities are retained in the claim qualification."},{"id":"evidence-prov-fsf-v2-0-government-disclosure","sourceRole":"primary","support":"supports","locator":"Governance and Accountability, “Disclosure”, first sentence: PDF p. 8 in the Internet Archive capture of 2025-02-05; PDF pp. 7–8 in the served file, where a page break and footnote 14 interrupt the sentence after “facilitate the”","passage":"If we assess that a model has reached a CCL that poses an unmitigated and material risk to overall public safety, we aim to share information with appropriate government authorities where it will facilitate the development of safe AI.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-03","note":"Verbatim excerpt checked by an automated research assistant on 2026-10-03 against the Internet Archive capture of 2025-02-05 (the edition as first published) and the current file, which carries a later documented correction; the passage is identical in both. Extraction note: Contiguous in the capture; in the served file the sentence runs across a page break, with footnote 14 and the page number between “facilitate the” and “development of safe AI.” The sentence’s words are identical in both copies. Human editorial review is pending."},{"id":"evidence-ts-c2-drexler-1986","sourceRole":"primary","support":"supports","locator":"Notes for chapter 5, note \"... social AI ...\"","passage":"a Turing-test passer would of necessity be designed to understand and deceive people","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-06","note":"Checked against the 1997 capture of the notes page."},{"id":"evidence-ts-a4-christiano-2018-takeoff-contests","sourceRole":"primary","support":"supports","locator":"Section \"Slow takeoff\", subsection \"Operationalizing slow takeoff\"","passage":"There will be a complete 4 year interval in which world output doubles, before the first 1 year interval in which world output doubles.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-06","note":"Checked against the current page on 2026-10-06."},{"id":"evidence-ts-a4-christiano-2018-takeoff-contests-2","sourceRole":"primary","support":"supports","locator":"Section \"Reasons to expect fast takeoff\", subsection \"Recursive self-improvement\"","passage":"Before there is AI that is great at self-improvement there will be AI that is mediocre at self-improvement. Powerful AI can be used to develop better AI (amongst other things). This will lead to runaway growth. This on its own is not an argument for discontinuity","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-06","note":"Checked against the current page on 2026-10-06."},{"id":"evidence-ts-a4-christiano-2022-context","sourceRole":"primary","support":"supports","locator":"Section \"Disagreements\", bullet beginning \"One important factor seems to be\"","passage":"AI systems avoid making major technical contributions, or revealing the extent of their capabilities, because they are lying in wait to cause trouble later. But [...] SGD will be aggressively selecting against any AI systems who don’t do impressive-looking stuff.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-28","note":"The same passage as the B1 and B2 records. On 2026-10-30, by the maintainer’s decision of that day, the quotation was corrected to the archive capture of 2022-06-19 (quotation marks and apostrophes as in the capture), which a cross-family archive check read on 2026-10-28; the proposition is unchanged."},{"id":"evidence-yk-t05-c","sourceRole":"primary","support":"supports","locator":"Thou Art Godshatter, paragraph beginning 'Yet when the blind idiot god'","passage":"Yet when the blind idiot god created protein computers, its monomaniacal focus on inclusive genetic fitness was not faithfully transmitted.  Its optimization criterion did not successfully quine.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-06","note":"Candidate pack passage T05-c. The quotation (27 words) matched the retrieved expression (SHA-256 b6abf133e881a827dca01e79a7c5493835c6d026c00fcdd6d46fd6b08086551d) when checked on 2026-10-06."},{"id":"evidence-yk-t05-e1","sourceRole":"primary","support":"supports","locator":"message body, paragraph beginning '\"Parenting\" isn't going to help'","passage":"\"Parenting\" isn't going to help … emotions that won't exist in an AI.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-07","note":"Candidate pack passage T05-e1. The quotation (13 words) matched the retrieved expression (SHA-256 3b2d6b2b2285f03c56382a096fc6ebdaf54003e6bac0124bf96ba0bf3665c7d2) when checked on 2026-10-07."},{"id":"evidence-yk-t07-e","sourceRole":"primary","support":"supports","locator":"TIME, 29 March 2023, paragraph beginning 'There’s no proposed plan for how we could do any such thing and survive'","passage":"OpenAI’s openly declared intention is to make some future AI do our AI alignment homework. … enough to get any sensible person to panic.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-06","note":"Candidate pack passage T07-e. The quotation (24 words; trimmed 2026-10-13 from 33 words) matched the retrieved expression (SHA-256 4a8c77e15456ca68b530df8bd825296e3923f621b0c0b09dacce312ae94182d6) when checked on 2026-10-06."},{"id":"evidence-aia-art101-gpai-fines-1","sourceRole":"primary","support":"supports","locator":"Article 101(1), opening","passage":"The Commission may impose on providers of general-purpose AI models fines not exceeding 3 % of their annual total worldwide turnover in the preceding financial year or EUR 15 000 000, whichever is higher","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. In the Official Journal text the sentence continues with a stray full stop and comma before “when the Commission finds”. 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-art101-gpai-fines-2","sourceRole":"primary","support":"supports","locator":"Article 101(1), points (a) to (d)","passage":"The Commission may fine a provider when it finds that the provider intentionally or negligently (a) infringed the relevant provisions, (b) failed to comply with a request for a document or information under Article 91 or supplied incorrect, incomplete or misleading information, (c) failed to comply with a measure requested under Article 93, or (d) failed to give the Commission access to the model for an evaluation under Article 92.","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-g02-g","sourceRole":"primary","support":"supports","locator":"Section 4 \"Some Variations\", learning-by-doing paragraph","passage":"Otherwise, we no longer have diminishing returns and steady growth solutions are no longer possible; our model predicts a continually accelerating growth rate.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-14","note":"Candidate pack passage G02-g (research/economics-growth). The quotation (23 words) matched the checked expression (hanson-aigrow) in the pack's verification. Re-checked at admission on 2026-10-14 against a copy retrieved that day.","contextOnly":{"reason":"Alone, it reads as the paper's headline prediction; the author presents it as a variant he does not adopt.","requires":"The paper's statement that it sticks with exogenous growth models \"to remain conservative\"."}},{"id":"evidence-ec-m07-01","sourceRole":"primary","support":"supports","locator":"pdf page 1; Abstract","passage":"we begin by implementing a novel methodology to estimate the probability of computerisation for 702 detailed occupations, using a Gaussian process classifier.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-14","note":"Candidate pack passage M07-01 (research/economics-mainstream). The quotation (22 words) matched the retrieved current copy (SHA-256 1d0ef561ebef9ececda63403a40860aa172be953fe2b050d27de8987c9e852fc) in the pack's verification. Re-checked at admission on 2026-10-14 against a copy retrieved that day."},{"id":"evidence-ec-m07-02","sourceRole":"primary","support":"supports","locator":"pdf page 44; Section VI, Conclusions (printed p. 44)","passage":"According to our estimates around 47 percent of total US employment is in the high risk category. We refer to these as jobs at risk – i.e. jobs we expect could be automated relatively soon, perhaps over the next decade or two.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-14","note":"Candidate pack passage M07-02 (research/economics-mainstream). The quotation (42 words) matched the retrieved current copy (SHA-256 1d0ef561ebef9ececda63403a40860aa172be953fe2b050d27de8987c9e852fc) in the pack's verification. Re-checked at admission on 2026-10-14 against a copy retrieved that day.","contextOnly":{"reason":"Misleads when shown alone (GPT cross-family check of 2026-10-12, accepted by the maintainer on 2026-10-12).","requires":"The preceding sentence explicitly disclaims estimating how many jobs will actually be automated."}},{"id":"evidence-ec-e01-28","sourceRole":"primary","support":"supports","locator":"pdf p.2; Abstract","passage":"Automating R&D (so that machines can self-improve) would accelerate the transformation, but may not produce it in isolation.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-16","note":"Candidate pack passage E01-28 (research/economics-mainstream/esoteric). The quotation (18 words) matched the retrieved current copy (SHA-256 06dda75180302720e0843a180c00d04fa99e38db0f5d6d4e06d7f29388a115bb) in the pack's verification of 2026-10-15. Re-checked at admission on 2026-10-16 against a copy retrieved that day.","contextOnly":{"reason":"A model scenario, not the author's own forecast (maintainer decision D3).","requires":"The model's conditions and the author's own assessment."}},{"id":"evidence-ec-e01-29","sourceRole":"primary","support":"supports","locator":"pdf p.29; §6 Conclusion, 'From AI to automation'","passage":"suggests that capital is, in some sense, contributing significantly to its own improvement. As AI R&D is automated ever more fully, AI capabilities could advance ever more rapidly.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-16","note":"Candidate pack passage E01-29 (research/economics-mainstream/esoteric). The quotation (28 words) matched the retrieved current copy (SHA-256 06dda75180302720e0843a180c00d04fa99e38db0f5d6d4e06d7f29388a115bb) in the pack's verification of 2026-10-15. Re-checked at admission on 2026-10-16 against a copy retrieved that day.","contextOnly":{"reason":"GPT cross-family check 2026-10-14, accepted 2026-10-15","requires":"Begins mid-sentence: the subject is the high capital share in AI R&D (Besiroglu et al., 2024); the following sentence calls the calibrated models of this feedback speculative in part."}},{"id":"evidence-ec-e01-31","sourceRole":"primary","support":"supports","locator":"pdf p.30; §6 Conclusion, 'Preparing for dangerous AI'","passage":"a “software-only singularity” driven purely by R&D automation presents unique challenges through its potential for rapid, unobservable acceleration beneath conventional economic indicators","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-16","note":"Candidate pack passage E01-31 (research/economics-mainstream/esoteric). The quotation (22 words) matched the retrieved current copy (SHA-256 06dda75180302720e0843a180c00d04fa99e38db0f5d6d4e06d7f29388a115bb) in the pack's verification of 2026-10-15. Re-checked at admission on 2026-10-16 against a copy retrieved that day."},{"id":"lg-ev-mx-aldahle-departure-2026","sourceRole":"primary","support":"supports","locator":"Post, opening lines","passage":"Al-Dahle writes that he is joining Airbnb as CTO and refers to his time at Meta.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The paraphrase’s verbatim anchor “Airbnb as CTO” matched the retrieved text on 2026-10-15 and again in a fresh retrieval the same day. Read through X’s own syndication endpoint for the post’s text, which truncates long notes. 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":"x_syndication"},{"id":"lg-ev-mx-aldahle-departure-2026-title","sourceRole":"primary","support":"context","locator":"Byline under the title (Meta newsroom, Brazilian Portuguese edition, 19 December 2024)","passage":"Ahmad Al-Dahle, VP e Head de Inteligência Artificial Generativa da Meta","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-goo-dragan-home","sourceRole":"primary","support":"supports","locator":"Home, paragraph 1","passage":"currently on leave to head AI Safety and Alignment at Google DeepMind","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. No human editorial review.","retrievalRoute":"direct"},{"id":"lg-ev-goo-dragan-byline-2026","sourceRole":"primary","support":"context","locator":"Byline","passage":"Anca Dragan VP, AI Safety & Behavior","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. No human editorial review.","retrievalRoute":"direct"},{"id":"lg-ev-mx-wu-cofounded-xai","sourceRole":"secondary","support":"supports","locator":"Paragraphs 2–3 (the sentence ending “now sits at 10” closes paragraph 2; the list of names opens paragraph 3)","passage":"the number of cofounder departures now sits at 10. Guodong Zhang, Zihang Dai, Toby Pohlen, Jimmy Ba, Tony Wu, and Greg Yang have all stepped away since January.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The quotation (28 words) matched the retrieved text on 2026-10-15 and again in a fresh retrieval the same day. Read directly from the publisher. Matching ignored differences of whitespace, case, quotation marks, apostrophes, dashes and Unicode compatibility forms, so the source’s typography can differ. A cross-family machine check (2026-10-16/17) re-read the record’s sources and agreed with the record. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-ap-marvin-minsky-minsky-1961-steps","sourceRole":"primary","support":"supports","locator":"Page head: author byline","passage":"Credited as “Marvin Minsky” (page head: author byline).","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-27","note":"Credited name checked by an automated research assistant on 2026-10-27 in a direct retrieval and checked by a reviewer from a different model family on 2026-10-27. Given as a labelled paraphrase because the source records no document length against which a quotation can be checked. Human editorial review is pending.","retrievalRoute":"direct"},{"id":"evidence-ap-eliezer-yudkowsky-lw-2008-what-i-think-if-not-why","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-fsf-2024-d1-persuasion","sourceRole":"primary","support":"supports","locator":"FSF · 2024: Whole text layer, PDF pp. 1–7 (all sections, both tables, footnotes and acknowledgements read)","passage":"Within the inspected scope, no passage meeting the comparison question “Persuasion or manipulation” was found.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-26","note":"Absence recorded by an automated research assistant on 2026-10-15 for the framework comparison, audited in a fresh context, checked by a reviewer from a different model family on 2026-10-15 and re-checked on 2026-10-26. Human editorial review is pending."},{"id":"ev-cmp-fsf-v2-0-d8-process-1","sourceRole":"primary","support":"supports","locator":"PDF p. 1 (printed p. 1, unnumbered), opening section, last paragraph, third sentence","passage":"The edition says Google DeepMind will review the Framework from time to time and expects it to change substantially.","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-faif-2025-d7-internal-1","sourceRole":"primary","support":"supports","locator":"PDF p. 8, 'Operational and Societal Risks' › '1. Public transparency and third-party review', numbered item 1 ('Frontier AI Framework adherence'), second sentence","passage":"The edition says xAI allows its employees to report concerns about non-adherence anonymously, with protection from retaliation.","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-faif-2025-d7-internal-2","sourceRole":"primary","support":"supports","locator":"PDF p. 9, 'Operational and Societal Risks' › '4. Governance Approach', first paragraph, second sentence","passage":"The edition says risk owners are responsible as well for periodic audits that enforce the framework's implementation.","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-faif-2026-d8-changes","sourceRole":"primary","support":"supports","locator":"FAIF · June 2026 edition: Whole text layer, PDF pp. 1–9 (p. 10 blank): all sections read","passage":"Within the inspected scope, no passage meeting the comparison question “Stated changes from the previous edition” was found.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-26","note":"Absence recorded by an automated research assistant on 2026-10-15 for the framework comparison, audited in a fresh context, checked by a reviewer from a different model family on 2026-10-15 and re-checked on 2026-10-26 against an Internet Archive capture, because the publisher copy was no longer served (research/archive-checks-2026-10-26/codex-report.md). Human editorial review is pending."},{"id":"c2-ev-bytedance-seed-coder-licence","sourceRole":"primary","support":"supports","locator":"Section 'License' and front-matter metadata","passage":"This project is licensed under the MIT License.","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 (8 words) matched the retrieved text on 2026-10-27 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 (2026-10-27, item K035) re-read the record’s sources and agreed with the record. No human editorial review.","retrievalRoute":"direct"},{"id":"c2-ev-bytedance-seed-coder-licence-meta","sourceRole":"primary","support":"supports","locator":"Front-matter metadata","passage":"license: mit","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 (2 words) matched the retrieved text on 2026-10-27 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 (2026-10-27, item K035) re-read the record’s sources and agreed with the record. No human editorial review.","retrievalRoute":"direct"},{"id":"c2-ev-bytedance-seed-coder-licence-earlier","sourceRole":"primary","support":"supports","locator":"Front-matter metadata of the revision of 30 April 2025","passage":"license: apache-2.0","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 (2 words) matched the retrieved text on 2026-10-27 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 (2026-10-27, item K035) re-read the record’s sources and agreed with the record. No human editorial review.","retrievalRoute":"direct"},{"id":"c2-ev-governance-tc260-asgf-v3-published-1","sourceRole":"primary","support":"supports","locator":"News item, first paragraph","passage":"9月14日，在2026年国家网络安全宣传周开幕式上 ... 发布《人工智能安全治理框架3.0》","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 (33 words, each CJK character counted as one) matched the retrieved text on 2026-10-27 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. Elided quotation: the ellipses mark omitted words. English machine translation, not verified by a person: “On 14 September, at the opening ceremony of the 2026 National Cybersecurity Awareness Week ... [the committee] released the AI Safety Governance Framework 3.0”. No human editorial review.","retrievalRoute":"direct"},{"id":"c2-ev-governance-tc260-asgf-v3-published-2","sourceRole":"primary","support":"supports","locator":"Preface (前言)","passage":"在《人工智能安全治理框架 1.0》（2024年）、《人工智能安全治理框架2.0》（2025年）基础上，研究编制《人工智能安全治理框架3.0》","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 (43 words, each CJK character counted as one) matched the retrieved text on 2026-10-27 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. English machine translation, not verified by a person: “on the basis of the AI Safety Governance Framework 1.0 (2024) and the AI Safety Governance Framework 2.0 (2025), researched and prepared the AI Safety Governance Framework 3.0”. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-fl-intl-bletchley-chair-external-evaluations-1","sourceRole":"primary","support":"supports","locator":"Third acknowledged point","passage":"Governments recognise their increasing role for seeing that external evaluations are undertaken for frontier AI models developed within their countries","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-28","note":"Candidate pack passage, frameworks to law (research/frameworks-to-law). The quotation (20 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. The point continues that the tests 'should be conducted before and after those models are deployed'. A cross-family machine check (2026-10-28, item 4) re-read the record’s sources and found full support. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-cpo-dc-aws-gov-50bn-1","sourceRole":"primary","support":"supports","locator":"paragraph 1","passage":"Today, Amazon announced an investment of up to $50 billion to expand AI and supercomputing capabilities for Amazon Web Services (AWS) U.S. government customers.","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 (SHA-256 c2c224718427c46f611b0503ff9c9a11ac3c1bcc5a202d45b6efd301e74f6e3a; 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. A cross-family machine check (2026-10-29/30, item 6) answered the record’s question blind from fresh retrievals and agreed with the record. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-cpo-dc-aws-gov-50bn-2","sourceRole":"primary","support":"supports","locator":"paragraph 1","passage":"This investment, set to break ground in 2026, will add nearly 1.3 gigawatts of AI and supercomputing capacity","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-30","note":"Candidate pack passage, compute, power and orbit (research/compute-power-orbit). The quotation (18 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 c2c224718427c46f611b0503ff9c9a11ac3c1bcc5a202d45b6efd301e74f6e3a; 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. A cross-family machine check (2026-10-29/30, item 6) answered the record’s question blind from fresh retrievals and agreed with the record. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-cpo-dc-stargate-milam-announced-1","sourceRole":"primary","support":"supports","locator":"paragraph 3","passage":"located in Milam County, Texas, and will be developed in partnership with SB Energy","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-30","note":"Candidate pack passage, compute, power and orbit (research/compute-power-orbit). The quotation (14 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 a3d77f8db28800481eac9625ed0caf759b0f12015cf3b6f9f139cd330cc28292; 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. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-cpo-dc-stargate-milam-announced-2","sourceRole":"primary","support":"supports","locator":"paragraph 3","passage":"the release says the two other sites will be developed through a partnership by SoftBank and OpenAI, and describes SB Energy as a SoftBank Group company providing powered infrastructure.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-30","note":"Candidate pack passage, compute, power and orbit (research/compute-power-orbit). The paraphrase is Selekoa’s, written from the text retrieved for the candidate pack on 2026-10-28; it is not a quotation and was not machine-matched. The cross-family check found it faithful. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-ab-ev-usaisi-claude35-tooling-pilot-1","sourceRole":"primary","support":"supports","locator":"'Biological Capabilities' section, key findings, second bullet","passage":"US AISI piloted an evaluation method that augments the AI model by providing it access to bioinformatic tools to assist in research task questions, which when tested using Sonnet 3.5 resulted in increased performance beyond the model alone, at times exceeding measured human expert baselines.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-30","note":"Candidate pack passage, AI and biosecurity (research/ai-bio). The quotation (45 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 8e9a0960ed43d37fa5d161337886069510323506e77889b1301eb294c16e4b90; 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. A cross-family machine check (2026-10-30, item 19) re-read the record’s sources and found full support. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-ab-ev-claude-sonnet45-bio-categories-1","sourceRole":"primary","support":"supports","locator":"PDF p. 126 (printed p. 125), §9.2 CBRN evaluations, 'ASL-3 threat model' subsection","passage":"We primarily focus on biological risks with the largest consequences, such as pandemics.","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. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-ab-ev-claude-sonnet45-bio-categories-2","sourceRole":"primary","support":"supports","locator":"PDF p. 126 (printed p. 125), §9.2 CBRN evaluations, the 'ASL-3 threat model' and 'ASL-4 threat model' subheadings","passage":"Section 9.2 is set out under two subheadings, 'ASL-3 threat model' and 'ASL-4 threat model', each stating a separate capability threshold for the CBRN domain.","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. Added after the audit of 2026-10-28: the statement's second clause rested on no cited passage, the first locator naming only the ASL-3 subsection. Explicit paraphrase; neither threshold's own wording is reproduced, under the hazard-safe rule. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-ab-ev-anthropic-hub-sonnet55-cb1-1","sourceRole":"primary","support":"supports","locator":"Claude Sonnet 5.5 section, 'Testing Methods and Results' and the chemical and biological summary","passage":"On chemical and biological weapons, we treat Claude Sonnet 5.5 as able to significantly help someone with a basic technical background make known chemical or biological weapons, which is our CB-1 threshold.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-30","note":"Candidate pack passage, AI and biosecurity (research/ai-bio). The quotation (32 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 180bb54e99e373f383f2d06ea3975f8819bd2708bc49a9f5fafd088074259a63; 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 same section states that the model was deployed with the protections used for Claude Opus 5. A cross-family machine check (2026-10-30, item 45) re-read the record’s sources and found full support. No human editorial review.","retrievalRoute":"direct"}],"at":{"entities":[1,2,3,5,19,22,50,56,57,60,63,157,188,189,264,269,327,334,339,348,354,356,383,401,429,444,491,505,520,524,563,585,601,615,628,643,835,901,994,1001,1004,1019,1049,1134,1175,1233,1237,1266],"claims":[121,379,520,812,951,975,1021,1027,1036,1205,1255,1440,1502,1834,1916,2013,2293,2448,2592,2610,2678,2692,2785,2846,2941,3125,3153,3371,3399,3403],"sources":[44,45,51,52,61,171,264,270,332,339,350,368,396,411,457,475,516,550,633,637,692,696,704,820,821,833,834,850,890,961,981,1082,1084],"evidence":[140,402,586,1022,1173,1174,1199,1248,1254,1263,1473,1474,1571,1790,1791,1873,1874,1875,2358,2359,2478,2479,2602,2940,3095,3256,3276,3359,3360,3380,3532,3533,3534,3644,3645,3800,4129,4130,4193,4194,4629,4676,4677,4682]},"version":"1lg47t7"}