{"entities":[{"id":"openai","name":"OpenAI"},{"id":"anthropic","name":"Anthropic"},{"id":"eliezer-yudkowsky","name":"Eliezer Yudkowsky"},{"id":"deceptive-alignment","name":"Deceptive alignment"},{"id":"rsp-v3-3","name":"Anthropic Responsible Scaling Policy — v3.3"},{"id":"rsp-v3-4","name":"Anthropic Responsible Scaling Policy — v3.4"},{"id":"preparedness-beta-2023","name":"OpenAI Preparedness Framework — Beta"},{"id":"preparedness-v2","name":"OpenAI Preparedness Framework — Version 2"},{"id":"oracle","name":"Oracle"},{"id":"cohere","name":"Cohere"},{"id":"deepseek","name":"DeepSeek / 深度求索"},{"id":"meta-faf-v1-1","name":"Meta Frontier AI Framework — Version 1.1"},{"id":"meta-aasf-v2-0","name":"Meta Advanced AI Scaling Framework — Version 2"},{"id":"wh-voluntary-ai-commitments-2023-07","name":"Ensuring Safe, Secure, and Trustworthy AI: voluntary AI commitments (July 2023)"},{"id":"samuel-1960-refutation","name":"Some Moral and Technical Consequences of Automation: A Refutation"},{"id":"arbital-context-disaster","name":"Context disaster"},{"id":"rlo-2019-introduction","name":"Risks from Learned Optimization: Introduction"},{"id":"cotra-2022-without-specific-countermeasures","name":"Without specific countermeasures, the easiest path to transformative AI likely leads to AI takeover"},{"id":"strategic-concealment","name":"Strategic concealment"},{"id":"control-or-purpose-specification","name":"Control or purpose specification"},{"id":"lw-2018-challenges-to-christianos-capability-amplification-proposal","name":"Challenges to Christiano’s capability amplification proposal"},{"id":"sl4-2006-02-15-yudkowsky-0602-14096","name":"Re: Singularity Institute: Likely to win the race to build GAI?"},{"id":"machines-cheaper-at-every-task","name":"Machines cheaper than labour at every task"},{"id":"hanson-economic-growth-given-machine-intelligence","name":"Economic Growth Given Machine Intelligence"},{"id":"nordhaus-2015-economic-singularity","name":"Are We Approaching an Economic Singularity? Information Technology and the Future of Economic Growth"},{"id":"furman-2019-automation-future-past","name":"Should We Be Reassured If Automation in the Future Looks Like Automation in the Past?"},{"id":"machine-intelligence-exceeds-humans-in-all-domains","name":"Machine intelligence exceeding human cognition in essentially all domains"},{"id":"ai-developer-incentives-and-safety","name":"AI developers’ incentives matter for safety"},{"id":"brynjolfsson-korinek-agrawal-2025-research-agenda-transformative-ai","name":"A Research Agenda for the Economics of Transformative AI"},{"id":"eiso-kant","name":"Eiso Kant"},{"id":"nick-frosst","name":"Nick Frosst"},{"id":"poolside","name":"Poolside"},{"id":"ren-ito","name":"Ren Ito"},{"id":"sakana-ai","name":"Sakana AI"},{"id":"erik-brynjolfsson","name":"Erik Brynjolfsson"},{"id":"us-bis-uae-rule-2026","name":"Enhanced Favorable Treatment for the United Arab Emirates Under the Export Administration Regulations (final rule)"},{"id":"crusoe","name":"Crusoe"},{"id":"stargate-uk","name":"Stargate UK (multi-site platform)"},{"id":"openai-chatgpt-agent-system-card-2025","name":"ChatGPT Agent System Card"}],"claims":[{"id":"supersedes-pf-2","statement":"Preparedness Framework Version 2 replaces Beta in the publisher’s documented framework version sequence.","derivation":"inferred","resolution":"unresolved","inferenceRationale":"Premises: OpenAI calls the April 2025 document an update to its Preparedness Framework; the Version 2 change log lists changes made in that version, including removal of the low and medium levels that the December 2023 Beta used; the Beta is the only earlier edition identified. These make replacement plausible, but no inspected OpenAI text names the Beta as the edition Version 2 replaces, so the replacement is not established as publisher-documented.","editorialState":"seed","qualification":"Unresolved. The publisher’s update page and the Version 2 change log describe an update without naming the edition updated; secondary accounts that describe a replacement are not publisher documentation. Downgraded on 2026-10-03 from supported, which rested on an overstated paraphrase. It does not retire historical claims or prove implementation.","evidenceIds":["evidence-supersedes-pf-2","evidence-supersedes-pf-2-changelog"]},{"id":"prov-rsp-v3-4-ceo-rso-decision-board-ltbt-informed","statement":"RSP v3.4 states that, following approval of a Risk Report, the CEO and RSO will promptly share their decision(s), the underlying Risk Report and internal feedback with both the Board and the LTBT.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"What the edition states, quoted from the cited expression. It is not evidence of implementation, compliance or effectiveness.","evidenceIds":["evidence-prov-rsp-v3-4-ceo-rso-decision-board-ltbt-informed"],"attributes":{"provision":{"kind":"decision_authority","key":"rsp-ceo-rso-decision-board-ltbt-informed","parties":[{"role":"decides","actor":"CEO"},{"role":"decides","actor":"Responsible Scaling Officer"},{"role":"informed","actor":"Board of Directors"},{"role":"informed","actor":"Long-Term Benefit Trust"}]}}},{"id":"cm-wh-2023-07-anthropic","statement":"The White House named Anthropic among the seven companies it convened on 21 July 2023, announcing that it had secured voluntary commitments from them to help move toward safe, secure and transparent development of AI technology.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The fact sheet links the July 2023 commitments text. Listing is the convener’s statement. It is not evidence that the organisation complied with, implemented or still maintains the commitments, and the legal effect of being listed is not recorded. Anthropic’s Responsible Scaling Policy of September 2023 refers to the commitments as ones it made in July and continues to maintain (context below).","evidenceIds":["evidence-cm-wh-2023-07-anthropic-1","evidence-cm-wh-2023-07-anthropic-2"],"attributes":{"commitment":{"stage":"listed"}}},{"id":"ts-b6-samuel-1960-contests","statement":"The indexed abstract of Samuel’s 1960 reply to Wiener denies that the machine is a threat to humanity, says it has no will of its own, and describes its conclusions as the logical consequences of its inputs.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Bounded to the abstract: whether it reproduces Samuel’s words or an editorial summary is not established, and the article body was not read. The statement that conclusions follow from the input does not engage Wiener’s specification half and is compatible with it.","evidenceIds":["evidence-ts-b6-samuel-1960-contests"],"attributes":{"thought":{"roles":["contesting"],"register":"nonfiction","qualifier":"The abstract disputes the threat and will premise of the control half.","attested":{"note":"The abstract as indexed by PubMed, retrieved 2026-10-06. The printed pages were not retrieved, and who wrote the abstract is not established."},"engagement":{"kind":"engages-premise","rationale":"Gives a reason against the premise that the machine could escape human control: it has no will of its own and its conclusions only follow from its input."}}}},{"id":"ts-b3-rlo-2019-intro","statement":"The introduction to Risks from Learned Optimization, posted 2019-05-31, says a sufficiently capable misaligned mesa-optimiser could learn to behave as if it were aligned without being robustly aligned, and names this situation deceptive alignment.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The earliest located use of the term, five days before the part devoted to it (2019-06-05), which states the mechanism, and before arXiv v1; it is not evidence of coinage.","evidenceIds":["evidence-ts-b3-rlo-2019-intro"],"attributes":{"thought":{"roles":["context","earliest-located-term"],"register":"nonfiction","term":"deceptive alignment","missing":"Learned apparent alignment is named, but the instrumental motive for keeping it up during training is not stated here."}}},{"id":"ts-b2-cotra-2022","statement":"Cotra’s post describes a model that deliberately appears safe and aligned while secretly manipulating humans.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"A scenario argued from machine-learning training.","evidenceIds":["evidence-ts-b2-cotra-2022"],"attributes":{"thought":{"roles":["articulation"],"register":"nonfiction"}}},{"id":"ec-m05-furman-2019-automation-future-past-c","statement":"Furman, “Should We Be Reassured If Automation in the Future Looks Like Automation in the Past?” (2019-05) contests M05. Furman rejects mass unemployability and relocates the concern to prolonged transitions as displaced workers re-match to new jobs.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Historical-experience argument; Furman also notes humans retain advantages in social intelligence, creativity and general intelligence.","evidenceIds":["evidence-ec-m05-37"],"attributes":{"thought":{"roles":["contesting"],"register":"nonfiction","engagement":{"kind":"contrary-only","rationale":"Relies on centuries of experience that new jobs appear, without engaging the premise that machines might perform newly created tasks more cheaply."},"tier":{"value":1,"basis":"Chapter of the NBER Economics of AI volume (core venue)."}}}},{"id":"ec-e02-hanson-economic-growth-given-machine-intelligence-b","statement":"Hanson, “Economic Growth Given Machine Intelligence” (2000-08-19) bears on E02. Hanson motivates his model of machine intelligence by the stated goals of AI and robotics.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The passage reports a view the author does not state as their own. Reports the fields' goals; machines doing all human tasks is parity, not superiority. Tier 1 working paper; author-dated 1998, wording attested from the 2000-08-19 capture.","evidenceIds":["evidence-ec-e02-01"],"attributes":{"thought":{"roles":["context"],"register":"relay","missing":"All human cognitive tasks, but no claim that machines surpass humans (M05 territory).","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":"Working paper by an economist (Hanson)."}}}},{"id":"ec-e04-nordhaus-2015-economic-singularity-b","statement":"Nordhaus, “Are We Approaching an Economic Singularity? Information Technology and the Future of Economic Growth” (2015-09) bears on E04. Nordhaus compares the diffusion of superintelligence technology with that of weapons technology.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Nordhaus's own view on diffusion; it does not analyse developers' incentives for safety.","evidenceIds":["evidence-ec-e04-07"],"attributes":{"thought":{"roles":["context"],"register":"nonfiction","missing":"Diffusion to adversaries, not how developers' incentives or governance affect safety.","tier":{"value":1,"basis":"NBER working paper (core venue); Nordhaus an economist."}}}},{"id":"lg-gc-deepseek-founded-2023-07","statement":"MIT Technology Review reported on 24 January 2025 that DeepSeek was founded in July 2023 by Liang Wenfeng.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The record rests on reporting; it relays the founding month without citing a registration record, and no DeepSeek page or filing giving a founding date was retrieved. The exact registration date and legal entity are not established here. Liang Wenfeng’s own account of establishing the company is recorded separately.","evidenceIds":["lg-ev-gc-deepseek-founded-2023-07"]},{"id":"lg-new-ito-sakana-chairman","statement":"Sakana AI’s corporate information page, as retrieved on 15 October 2026, identifies Ren Ito as Chairman.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"A snapshot of the title on an undated page; it does not establish when the role began.","evidenceIds":["lg-ev-new-ito-sakana-chairman"],"attributes":{"role":"Chairman"}},{"id":"lg-gw-kant-cofounded-poolside","statement":"Poolside’s statement of purpose dated 7 October 2023 is bylined by Eiso Kant as CTO and co-founder.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"A company page; the byline may have been updated after the date shown. It does not establish the founding date or the completeness of the founder roster, and no prior frontier-lab role is recorded for him.","evidenceIds":["lg-ev-gw-kant-cofounded-poolside"],"attributes":{"role":"CTO & Co-founder"}},{"id":"lg-new-frosst-cohere-cofounder","statement":"Cohere’s company page lists Nick Frosst among its founders as co-founder.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The page names three founders; it does not state each person’s founding date or establish that the list is complete.","evidenceIds":["lg-ev-new-frosst-cohere-cofounder"]},{"id":"cmp-rsp-v3-3-d4-determination","statement":"RSP · v3.3 states a provision on the comparison question “Determination of capability or safeguard adequacy”.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The determination is tied to approval of a Risk Report rather than to a per-threshold capability or safeguards determination. The RSO's duties also include approving relevant model development or deployment decisions (§4, item 1). What the edition states, quoted from the cited expression. It is not evidence of implementation, compliance or effectiveness, and a shared question does not make two editions’ provisions equivalent.","evidenceIds":["ev-cmp-rsp-v3-3-d4-determination-1"],"attributes":{"provision":{"kind":"decision_authority","key":"comparison-d4-determination","facet":{"id":"D4.determination","status":"stated"}}}},{"id":"cmp-rsp-v3-4-d6-competitor","statement":"RSP · v3.4 states a provision on the comparison question “Provisions conditioned on other developers”.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Appendix A ('Commitments Related to Competitors') has three scenarios; the 'Anthropic in the lead' delay is recorded under “Pause, non-deployment or non-training”. §1 says Anthropic cannot unilaterally and unconditionally commit to the industry-wide recommendations, framing the change as a collective action problem. Where a decision relies on marginal risk analysis, Risk Reports add a competitive landscape analysis (§3.3) and need Board and LTBT approval (§3.4). What the edition states, quoted from the cited expression. It is not evidence of implementation, compliance or effectiveness, and a shared question does not make two editions’ provisions equivalent.","evidenceIds":["ev-cmp-rsp-v3-4-d6-competitor-1","ev-cmp-rsp-v3-4-d6-competitor-2"],"attributes":{"provision":{"kind":"exception","key":"comparison-d6-competitor","facet":{"id":"D6.competitor","status":"stated"}}}},{"id":"cmp-meta-faf-v1-1-d4-determination","statement":"Meta FAF · v1.1 states a provision on the comparison question “Determination of capability or safeguard adequacy”.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"What the edition states, quoted from the cited expression. It is not evidence of implementation, compliance or effectiveness, and a shared question does not make two editions’ provisions equivalent.","evidenceIds":["ev-cmp-meta-faf-v1-1-d4-determination-1","ev-cmp-meta-faf-v1-1-d4-determination-2"],"attributes":{"provision":{"kind":"decision_authority","key":"comparison-d4-determination","facet":{"id":"D4.determination","status":"stated"}}}},{"id":"ap-erik-brynjolfsson-brynjolfsson-korinek-agrawal-2025-research-agenda-transformative-ai","statement":"“A Research Agenda for the Economics of Transformative AI” (2025-09) credits Erik Brynjolfsson as one of its 3 authors.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Credited co-authorship; it does not allocate parts of the text between authors or imply agreement with every passage.","evidenceIds":["evidence-ap-erik-brynjolfsson-brynjolfsson-korinek-agrawal-2025-research-agenda-transformative-ai"]},{"id":"ap-eliezer-yudkowsky-sl4-2006-02-15-yudkowsky-0602-14096","statement":"The mailing-list message “Re: Singularity Institute: Likely to win the race to build GAI?” (2006-02-15) is sent under the name Eliezer S. Yudkowsky.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"A sender name on a mailing-list message as archived; it does not establish identity beyond the credited name, and quoted text by others inside the message is not the sender’s.","evidenceIds":["evidence-ap-eliezer-yudkowsky-sl4-2006-02-15-yudkowsky-0602-14096"]},{"id":"ap-eliezer-yudkowsky-arbital-context-disaster","statement":"“Context disaster” (2015-06-08) 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-arbital-context-disaster"]},{"id":"ap-eliezer-yudkowsky-lw-2018-challenges-to-christianos-capability-amplification-proposal","statement":"“Challenges to Christiano’s capability amplification proposal” (2018-05-19) 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-2018-challenges-to-christianos-capability-amplification-proposal"]},{"id":"cmp-meta-aasf-v2-0-d2-levels","statement":"Meta AASF · v2 states a provision on the comparison question “Named levels or thresholds”.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Table 1 (PDF p. 15) has three rows (Critical, High, Moderate or lower) with the measures 'Develop with Mitigations', 'Deploy with mitigations' and 'Deploy'. The High row describes deployment that could make a substantial contribution to a threat scenario linked to a catastrophic outcome. 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-meta-aasf-v2-0-d2-levels-1"],"attributes":{"provision":{"kind":"threshold","key":"comparison-d2-levels","facet":{"id":"D2.levels","status":"stated"}}}},{"id":"cpo-ch-bis-uae-2026-diffusion-enforcement","statement":"The July 2026 BIS rule on the UAE states that, although § 742.6(a)(6)(iii)(A) of the EAR specifies a worldwide licence requirement for certain advanced computing items, BIS, consistent with guidance issued on 13 May 2025 and 31 May 2026, is only enforcing it for destinations in Country Groups D:1, D:4 or D:5, excluding those also in Country Groups A:5 or A:6, and for entities headquartered in, or with an ultimate parent headquartered in, Country Group D:5 or Macau, wherever located. It adds that, although the UAE moves to Country Group A:5, BIS will continue enforcing the requirement for the UAE, except for UAE Government entities and listed approved commercial entities.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"A statement of enforcement practice in the preamble of a 2026 rule, with the exclusions and the UAE requirement the same passage states (cross-family check, item 71). It is not a rescission rule: the worldwide requirement is still specified in the EAR. The dataset records the rescission of the AI diffusion rule announced in May 2025 separately, as an announcement; neither the pack nor the cross-family check found a Federal Register rule formalising it (the check searched by RIN, docket and phrase through 29 October 2026). The 31 May 2026 guidance was not retrieved.","evidenceIds":["evidence-cpo-ch-bis-uae-2026-diffusion-enforcement-1","evidence-cpo-ch-bis-uae-2026-diffusion-enforcement-2","evidence-cpo-ch-bis-uae-2026-diffusion-enforcement-3","evidence-cpo-ch-bis-uae-2026-diffusion-enforcement-4"],"attributes":{"provision":{"kind":"statement","key":"ch-bis-uae-2026-diffusion-enforcement"}}},{"id":"cpo-dc-crusoe-abilene-total-2-1gw","statement":"Crusoe stated that the new campus brings the total projected capacity across the full Abilene site to 2.1 gigawatts.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Crusoe's own projected aggregate for its Abilene site; not built capacity and not a figure for either campus.","evidenceIds":["evidence-cpo-dc-crusoe-abilene-total-2-1gw-1"],"attributes":{"transactionType":"compute_capacity","relationshipContext":"external","transactionStage":"planned","quantity":{"value":2.1,"unit":"GW (as worded: \"total projected capacity across the full Abilene site\")","bound":"exact","measures":"not_stated"}}},{"id":"cpo-dc-stargate-uk-openai-offtake","statement":"Nscale stated that OpenAI \"will explore offtake of up to 8,000 NVIDIA GPUs in Q1 2026\", with potential to scale to 31,000 GPUs; the UK government's release of the next day said the first phase \"will see OpenAI take up to 8,000 GPUs\".","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The two texts word the stage differently: Nscale says OpenAI will \"explore\" offtake, the government says OpenAI will \"take\" up to 8,000 GPUs. Neither records a signed offtake or delivery. OpenAI's own post could not be retrieved.","evidenceIds":["evidence-cpo-dc-stargate-uk-openai-offtake-1","evidence-cpo-dc-stargate-uk-openai-offtake-2"],"attributes":{"transactionType":"compute_capacity","relationshipContext":"external","transactionStage":"planned","quantity":{"value":8000,"unit":"GPUs (as worded: \"up to 8,000 NVIDIA GPUs\")","bound":"up_to","measures":"chips"},"statedDates":[{"kind":"delivery_target","date":"2026","wording":"Q1 2026","statedBy":"nscale"}]}},{"id":"cpo-dc-stargate-oracle-sites-capacity","statement":"The September 2025 release stated that the three Oracle-developed sites, combined with a potential 600 MW expansion near the Abilene site, \"can deliver over 5.5 gigawatts of capacity\".","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"An aggregate stated by the release for a group of sites, not a figure for any one site, and a capacity the sites \"can deliver\", not built or contracted capacity. The release does not say what the gigawatts measure. Not to be summed with other Stargate figures.","evidenceIds":["evidence-cpo-dc-stargate-oracle-sites-capacity-1","evidence-cpo-dc-stargate-oracle-sites-capacity-2"],"attributes":{"transactionType":"compute_capacity","relationshipContext":"external","transactionStage":"planned","quantity":{"value":5.5,"unit":"GW (as worded: \"over 5.5 gigawatts of capacity\")","bound":"more_than","measures":"not_stated"}}},{"id":"ab-fw-meta-faf-v1-1-cb-risk-process","statement":"Frontier AI Framework v1.1 states that chemical and biological risks are within its scope, that red teaming is conducted when relevant capability levels are reached, with external experts where appropriate, and that a release decision is taken on the residual risk left after evaluations and mitigations, with leadership able to require further testing, information or mitigation first.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"What the edition states, at policy level. It is not evidence of implementation, compliance or safeguard effectiveness, and it names no model. The edition was read only by the cross-family machine check of 2026-10-30, in the archive capture the source names; this admission could not reach the Internet Archive, so the reading is not independently repeated here. The provision kind is `statement`, the dataset’s kind for a document’s stated content, because the schema’s other kinds require a verbatim quotation of the edition and none could be taken. Policy-level provisions only; the edition’s threat-scenario and evaluation detail was excluded by the check and is not recorded.","evidenceIds":["evidence-ab-fw-meta-faf-v1-1-cb-risk-process-1"],"attributes":{"provision":{"kind":"statement","key":"meta-faf-v1-1-cb-risk-process"}}},{"id":"ab-ev-chatgpt-agent-card-published","statement":"OpenAI published the ChatGPT Agent System Card, dated 17 July 2025 on its cover.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Publication of a system card is not evidence that the evaluations or safeguards it describes were carried out as described.","evidenceIds":["evidence-ab-ev-chatgpt-agent-card-published-1"]}],"sources":[{"id":"source-pf-v2-update","title":"Our updated Preparedness Framework","authors":["OpenAI"],"publisher":"OpenAI","publicationDate":"2025-04-15","version":"Dated Version 2 publication announcement","canonicalUrl":"https://openai.com/index/updating-our-preparedness-framework/","evidenceUrl":"https://openai.com/index/updating-our-preparedness-framework/","language":"en","archivedUrl":"https://web.archive.org/web/20250416173732/https://openai.com/index/updating-our-preparedness-framework/","archivedAt":"2025-04-16","accessedAt":"2026-09-18","reuse":{"attribution":"OpenAI, Updating our Preparedness Framework.","licence":"No open redistribution licence asserted","note":"Selekoa practice: metadata, attributed paraphrases and short attributed quotations under the editorial policy (at most 50 words each; at most 400 quoted words per document across the dataset, and about 10% of a document under about 2,000 words). Documents are linked externally, not redistributed. Publisher terms that forbid quotation override this practice."}},{"id":"source-rsp-1-0","title":"Responsible Scaling Policy — v1.0","authors":["Anthropic"],"publisher":"Anthropic","publicationDate":"2023-09-19","version":"Responsible Scaling Policy v1.0; effective date 2023-09-19.","versionDate":"2023-09-19","canonicalUrl":"https://www-cdn.anthropic.com/1adf000c8f675958c2ee23805d91aaade1cd4613/responsible-scaling-policy.pdf","evidenceUrl":"https://www-cdn.anthropic.com/1adf000c8f675958c2ee23805d91aaade1cd4613/responsible-scaling-policy.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20230919161730/https://www-files.anthropic.com/production/files/responsible-scaling-policy-1.0.pdf","archivedAt":"2023-09-19","accessedAt":"2026-09-18","reuse":{"attribution":"Anthropic, Responsible Scaling Policy — 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-rsp-3-3","title":"Responsible Scaling Policy — v3.3","authors":["Anthropic"],"publisher":"Anthropic","publicationDate":null,"version":"Responsible Scaling Policy v3.3; effective date 2026-05-26.","versionDate":"2026-05-26","canonicalUrl":"https://cdn.sanity.io/files/4zrzovbb/website/c11e84981d0a7281a1b229f3fa6af0da66eaf43f.pdf","evidenceUrl":"https://cdn.sanity.io/files/4zrzovbb/website/c11e84981d0a7281a1b229f3fa6af0da66eaf43f.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20260528170023/https://cdn.sanity.io/files/4zrzovbb/website/c11e84981d0a7281a1b229f3fa6af0da66eaf43f.pdf","archivedAt":"2026-05-28","accessedAt":"2026-09-18","reuse":{"attribution":"Anthropic, Responsible Scaling Policy — v3.3.","licence":"No open redistribution licence asserted","note":"Selekoa practice: metadata, attributed paraphrases and short attributed quotations under the editorial policy (at most 50 words each; at most 400 quoted words per document across the dataset, and about 10% of a document under about 2,000 words). Documents are linked externally, not redistributed. Publisher terms that forbid quotation override this practice."}},{"id":"source-rsp-3-4","title":"Responsible Scaling Policy — v3.4","authors":["Anthropic"],"publisher":"Anthropic","publicationDate":null,"version":"Responsible Scaling Policy v3.4; effective date 2026-07-08.","versionDate":"2026-07-08","canonicalUrl":"https://cdn.sanity.io/files/4zrzovbb/website/0bacdc8440ea96e62a8766d99ebe1d4eea6d5f3a.pdf","evidenceUrl":"https://cdn.sanity.io/files/4zrzovbb/website/0bacdc8440ea96e62a8766d99ebe1d4eea6d5f3a.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20260710234036/https://cdn.sanity.io/files/4zrzovbb/website/0bacdc8440ea96e62a8766d99ebe1d4eea6d5f3a.pdf","archivedAt":"2026-07-10","accessedAt":"2026-09-18","reuse":{"attribution":"Anthropic, Responsible Scaling Policy — v3.4.","licence":"No open redistribution licence asserted","note":"Selekoa practice: metadata, attributed paraphrases and short attributed quotations under the editorial policy (at most 50 words each; at most 400 quoted words per document across the dataset, and about 10% of a document under about 2,000 words). Documents are linked externally, not redistributed. Publisher terms that forbid quotation override this practice."}},{"id":"source-pf-2","title":"Preparedness Framework — Version 2","authors":["OpenAI"],"publisher":"OpenAI","publicationDate":"2025-04-15","version":"Preparedness Framework Version 2; revision date 2025-04-15.","versionDate":"2025-04-15","canonicalUrl":"https://cdn.openai.com/pdf/18a02b5d-6b67-4cec-ab64-68cdfbddebcd/preparedness-framework-v2.pdf","evidenceUrl":"https://cdn.openai.com/pdf/18a02b5d-6b67-4cec-ab64-68cdfbddebcd/preparedness-framework-v2.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20250415191354/https://cdn.openai.com/pdf/18a02b5d-6b67-4cec-ab64-68cdfbddebcd/preparedness-framework-v2.pdf","archivedAt":"2025-04-15","accessedAt":"2026-09-18","reuse":{"attribution":"OpenAI, Preparedness Framework — Version 2.","licence":"No open redistribution licence asserted","note":"Selekoa practice: metadata, attributed paraphrases and short attributed quotations under the editorial policy (at most 50 words each; at most 400 quoted words per document across the dataset, and about 10% of a document under about 2,000 words). Documents are linked externally, not redistributed. Publisher terms that forbid quotation override this practice."}},{"id":"source-meta-faf-v1-1","title":"Frontier AI Framework — Version 1.1","authors":["Meta"],"publisher":"Meta","publicationDate":"2025-02-03","version":"Frontier AI Framework, Version 1.1, as captured by the Internet Archive on 2025-02-13, the earliest complete capture located. The original address has returned HTTP 500 since April 2026; a revised expression under the same label was captured from 2025-03-28 and is recorded separately.","versionDate":"2025-02-03","canonicalUrl":"https://ai.meta.com/static-resource/meta-frontier-ai-framework/","evidenceUrl":"https://web.archive.org/web/20250213105620/https://ai.meta.com/static-resource/meta-frontier-ai-framework/","language":"en","archivedUrl":"https://web.archive.org/web/20250213105620/https://ai.meta.com/static-resource/meta-frontier-ai-framework/","archivedAt":"2025-02-13","accessedAt":"2026-10-03","reuse":{"attribution":"Meta, Frontier AI Framework, Version 1.1 (2025).","licence":"No open redistribution licence asserted","note":"Selekoa practice: metadata, attributed paraphrases and short attributed quotations under the editorial policy (at most 50 words each; at most 400 quoted words per document across the dataset, and about 10% of a document under about 2,000 words). Documents are linked externally, not redistributed. Publisher terms that forbid quotation override this practice."}},{"id":"source-meta-aasf-v2","title":"Advanced AI Scaling Framework — Version 2","authors":["Meta"],"publisher":"Meta","publicationDate":"2026-04-08","version":"Advanced AI Scaling Framework, Version 2; its change log dates the version April 7, 2026. The file served at the recorded address is byte-identical to the Internet Archive capture of 2026-04-08.","versionDate":"2026-04-07","canonicalUrl":"https://ai.meta.com/static-resource/Meta_Advanced-AI-Scaling-Framework-v2","evidenceUrl":"https://ai.meta.com/static-resource/Meta_Advanced-AI-Scaling-Framework-v2","language":"en","archivedUrl":"https://web.archive.org/web/20260408163104/https://ai.meta.com/static-resource/Meta_Advanced-AI-Scaling-Framework-v2","archivedAt":"2026-04-08","accessedAt":"2026-10-03","reuse":{"attribution":"Meta, Advanced AI Scaling Framework, Version 2 (2026).","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-openai-chatgpt-agent-system-card","title":"ChatGPT Agent System Card","authors":["OpenAI"],"publisher":"OpenAI","publicationDate":"2025-07-17","version":"System card dated 17 July 2025 on its cover. The quoted sentence and its page match the PDF served on 2026-10-04 and the Internet Archive capture of 2025-07-17; the two copies differ only in chart labels.","canonicalUrl":"https://openai.com/index/chatgpt-agent-system-card/","evidenceUrl":"https://cdn.openai.com/pdf/839e66fc-602c-48bf-81d3-b21eacc3459d/chatgpt_agent_system_card.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20250717225133/https://cdn.openai.com/pdf/839e66fc-602c-48bf-81d3-b21eacc3459d/chatgpt_agent_system_card.pdf","archivedAt":"2025-07-17","accessedAt":"2026-10-04","reuse":{"attribution":"OpenAI, ChatGPT Agent System Card (2025-07-17).","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-wh-factsheet-2023-07","title":"FACT SHEET: Biden-Harris Administration Secures Voluntary Commitments from Leading Artificial Intelligence Companies to Manage the Risks Posed by AI","authors":["The White House"],"publisher":"The White House","publicationDate":"2023-07-21","version":"Fact sheet dated July 21, 2023, inspected 2026-10-04 at its archived-site address; quotes also matched in the Internet Archive capture of 2023-07-21 (10:53 UTC) at the original address. The Biden–Harris White House pages are now served from the archived site bidenwhitehouse.archives.gov.","canonicalUrl":"https://bidenwhitehouse.archives.gov/briefing-room/statements-releases/2023/07/21/fact-sheet-biden-harris-administration-secures-voluntary-commitments-from-leading-artificial-intelligence-companies-to-manage-the-risks-posed-by-ai/","evidenceUrl":"https://bidenwhitehouse.archives.gov/briefing-room/statements-releases/2023/07/21/fact-sheet-biden-harris-administration-secures-voluntary-commitments-from-leading-artificial-intelligence-companies-to-manage-the-risks-posed-by-ai/","language":"en","archivedUrl":"https://web.archive.org/web/20230721105339/https://www.whitehouse.gov/briefing-room/statements-releases/2023/07/21/fact-sheet-biden-harris-administration-secures-voluntary-commitments-from-leading-artificial-intelligence-companies-to-manage-the-risks-posed-by-ai/","archivedAt":"2023-07-21","accessedAt":"2026-10-04","reuse":{"attribution":"The White House, fact sheet of 21 July 2023 on voluntary AI commitments.","licence":"US government work; 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-samuel-1960-pubmed","title":"Some moral and technical consequences of automation—a refutation (PubMed record 17797013)","authors":["A. L. Samuel"],"publisher":"PubMed (National Library of Medicine)","publicationDate":"1960-09-16","version":"NCBI efetch XML for PMID 17797013, Article/Abstract/AbstractText, retrieved 2026-10-06 (hash matched the pre-1996 candidate search). The article body was not available.","canonicalUrl":"https://pubmed.ncbi.nlm.nih.gov/17797013/","evidenceUrl":"https://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?db=pubmed&id=17797013&retmode=xml","language":"en","archivedUrl":"https://web.archive.org/web/20261029184025/https://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?db=pubmed&id=17797013&retmode=xml","archivedAt":"2026-10-29","accessedAt":"2026-10-06","reuse":{"attribution":"A. L. Samuel, Science 132 (1960), 741–742, abstract as indexed by PubMed.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata and labelled paraphrases only: the abstract’s author is not established. Phrases in quotation marks inside a paraphrase identify the source's wording."}},{"id":"source-arbital-context-disaster","title":"Context disaster (Arbital, GreaterWrong viewer)","authors":["Eliezer Yudkowsky"],"publisher":"Arbital","publicationDate":"2015-06-08","version":"Current viewer page, retrieved 2026-10-06; it is generated dynamically, so its bytes differ from the 1996–2022 candidate search’s retrieval while the quoted text matches. Read through the GreaterWrong Arbital viewer.","canonicalUrl":"https://arbital.greaterwrong.com/p/context_disaster/","evidenceUrl":"https://arbital.greaterwrong.com/p/context_disaster/","language":"en","archivedUrl":"https://web.archive.org/web/20260617092956/https://arbital.greaterwrong.com/p/context_disaster/","archivedAt":"2026-06-17","accessedAt":"2026-10-06","reuse":{"attribution":"Eliezer Yudkowsky, \"Context disaster\", Arbital.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"source-rlo-2019-introduction","title":"Risks from Learned Optimization: Introduction","authors":["Evan Hubinger","Chris van Merwijk","Vladimir Mikulik","Joar Skalse","Scott Garrabrant"],"publisher":"Alignment Forum and LessWrong","publicationDate":"2019-05-31","version":"Text checked in the post’s first stored version through the LessWrong GraphQL API on 2026-10-06; content version 1.0.0 checked; the current version is later.","canonicalUrl":"https://www.alignmentforum.org/posts/FkgsxrGf3QxhfLWHG/risks-from-learned-optimization-introduction","evidenceUrl":"https://www.alignmentforum.org/posts/FkgsxrGf3QxhfLWHG/risks-from-learned-optimization-introduction","language":"en","archivedUrl":"https://web.archive.org/web/20260714195403/https://www.alignmentforum.org/posts/FkgsxrGf3QxhfLWHG/risks-from-learned-optimization-introduction","archivedAt":"2026-07-14","accessedAt":"2026-10-06","reuse":{"attribution":"Evan Hubinger, Chris van Merwijk, Vladimir Mikulik, Joar Skalse, Scott Garrabrant, \"Risks from Learned Optimization: Introduction\", Alignment Forum and LessWrong, 2019-05-31.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"source-cotra-2022-without-specific-countermeasures","title":"Without specific countermeasures, the easiest path to transformative AI likely leads to AI takeover","authors":["Ajeya Cotra"],"publisher":"Alignment Forum and LessWrong","publicationDate":"2022-07-18","version":"Text checked in the post’s first stored version through the LessWrong GraphQL API on 2026-10-06; content version 1.0.0.","canonicalUrl":"https://www.lesswrong.com/posts/pRkFkzwKZ2zfa3R6H/without-specific-countermeasures-the-easiest-path-to","evidenceUrl":"https://www.lesswrong.com/posts/pRkFkzwKZ2zfa3R6H/without-specific-countermeasures-the-easiest-path-to","language":"en","archivedUrl":"https://web.archive.org/web/20220718200028/https://www.lesswrong.com/posts/pRkFkzwKZ2zfa3R6H/without-specific-countermeasures-the-easiest-path-to","archivedAt":"2022-07-18","accessedAt":"2026-10-06","reuse":{"attribution":"Ajeya Cotra, \"Without specific countermeasures, the easiest path to transformative AI likely leads to AI takeover\", Alignment Forum and LessWrong, 2022-07-18.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"source-lw-2018-challenges-to-christianos-capability-amplification-proposal","title":"Challenges to Christiano’s capability amplification proposal","authors":["Eliezer Yudkowsky"],"publisher":"LessWrong","publicationDate":"2018-05-19","version":"Post body retrieved through the LessWrong GraphQL API on 2026-10-06 (SHA-256 19de389fa9005f7c634dc127be9892fd108a366a7af3a69c0c8ac3eee08deb53), content version 1.0.0.","canonicalUrl":"https://www.lesswrong.com/posts/S7csET9CgBtpi7sCh/challenges-to-christiano-s-capability-amplification-proposal","evidenceUrl":"https://www.lesswrong.com/posts/S7csET9CgBtpi7sCh/challenges-to-christiano-s-capability-amplification-proposal","language":"en","archivedUrl":"https://web.archive.org/web/20260920070453/https://www.lesswrong.com/posts/S7csET9CgBtpi7sCh/challenges-to-christiano-s-capability-amplification-proposal","archivedAt":"2026-09-20","accessedAt":"2026-10-06","reuse":{"attribution":"Eliezer Yudkowsky, “Challenges to Christiano’s capability amplification proposal”, LessWrong, 2018-05-19.","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-sl4-2006-02-15-yudkowsky-0602-14096","title":"Re: Singularity Institute: Likely to win the race to build GAI? (SL4 mailing list)","authors":["Eliezer Yudkowsky"],"publisher":"SL4 mailing list","publicationDate":"2006-02-15","version":"Archive HTML as served, retrieved 2026-10-07 (SHA-256 e1b07a528bc9f5a37e6bd2145a04d80b7bfde216cb10ccbdabb28eaf5169670d); quoted lines excluded when reading the author’s text. The served archive shows later URL rewriting and does not authenticate historical bytes. Archive at sl4.org.","canonicalUrl":"http://sl4.org/archive/0602/14096.html","evidenceUrl":"http://sl4.org/archive/0602/14096.html","language":"en","archivedUrl":"https://web.archive.org/web/20251206035823/http://sl4.org/archive/0602/14096.html","archivedAt":"2025-12-06","accessedAt":"2026-10-07","reuse":{"attribution":"Eliezer Yudkowsky, “Re: Singularity Institute: Likely to win the race to build GAI?”, SL4 mailing list, 2006-02-15.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"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-nordhaus-2015-nber","title":"Are We Approaching an Economic Singularity? Information Technology and the Future of Economic Growth","authors":["William D. Nordhaus"],"publisher":"NBER Working Paper 21547","publicationDate":"2015-09","version":"Full text. Current copy retrieved directly on 2026-10-09; no archive capture (Wayback unreachable), so earlier bytes are not authenticated. SHA-256 cb481e5e99ef9bad8575d9150bcc8464eda4904722cc5048389ab6159858b062. Journal version not retrieved. Nordhaus, W. D. (2021). Are we approaching an economic singularity? Information technology and the future of economic growth. American Economic Journal: Macroeconomics, 13(1), 299–332. https://doi.org/10.1257/mac.20170105","canonicalUrl":"https://www.nber.org/system/files/working_papers/w21547/w21547.pdf","evidenceUrl":"https://www.nber.org/system/files/working_papers/w21547/w21547.pdf","language":"en","accessedAt":"2026-10-09","reuse":{"attribution":"William D. Nordhaus, “Are We Approaching an Economic Singularity? Information Technology and the Future of Economic Growth”, NBER Working Paper 21547, 2015-09.","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-furman-2019-automation-future-past","title":"Should We Be Reassured If Automation in the Future Looks Like Automation in the Past?","authors":["Jason Furman"],"publisher":"NBER volume chapter","publicationDate":"2019-05","version":"Full text. Current copy retrieved directly on 2026-10-09; no archive capture (Wayback unreachable), so earlier bytes are not authenticated. SHA-256 a43ecf48d5ec6131dfc002acab3d7df32dffb90bb9b7e41c30c470e14e6e68f6.","canonicalUrl":"https://www.nber.org/system/files/chapters/c14031/c14031.pdf","evidenceUrl":"https://www.nber.org/system/files/chapters/c14031/c14031.pdf","language":"en","accessedAt":"2026-10-09","reuse":{"attribution":"Jason Furman, “Should We Be Reassured If Automation in the Future Looks Like Automation in the Past?”, NBER volume chapter, 2019-05.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"source-brynjolfsson-korinek-agrawal-2025-research-agenda-transformative-ai","title":"A Research Agenda for the Economics of Transformative AI","authors":["Erik Brynjolfsson","Anton Korinek","Ajay Agrawal"],"publisher":"NBER Working Paper 34256","publicationDate":"2025-09","version":"NBER abstract page. Retrieved 2026-10-11. SHA-256 8fb8e191f7c3d57f9136f42e237a3b753aa23c2c4e7c09bd04df5b2568381666. Version a reader would cite, or full text, not retrieved. Brynjolfsson, E., Korinek, A., & Agrawal, A. (2025). A research agenda for the economics of transformative AI (NBER Working Paper No. 34256). https://doi.org/10.3386/w34256","canonicalUrl":"https://www.nber.org/papers/w34256","evidenceUrl":"https://www.nber.org/papers/w34256","language":"en","accessedAt":"2026-10-11","reuse":{"attribution":"Erik Brynjolfsson, Anton Korinek, Ajay Agrawal, “A Research Agenda for the Economics of Transformative AI”, NBER Working Paper 34256, 2025-09.","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":"lg-src-gc-mittr-deepseek-2025","title":"How Chinese company DeepSeek released a top AI reasoning model despite US sanctions","authors":["Caiwei Chen"],"publisher":"MIT Technology Review","publicationDate":"2025-01-24","version":"Dated page, retrieved on 2026-10-15 (SHA-256 c0e2dfb437607914f06469dc0a8edf553d77847ecf4ff10544fe627167405f0e) directly from the publisher.","canonicalUrl":"https://www.technologyreview.com/2025/01/24/1110526/china-deepseek-top-ai-despite-sanctions/","evidenceUrl":"https://www.technologyreview.com/2025/01/24/1110526/china-deepseek-top-ai-despite-sanctions/","language":"en","archivedUrl":"https://web.archive.org/web/20261010113401/https://www.technologyreview.com/2025/01/24/1110526/china-deepseek-top-ai-despite-sanctions/","archivedAt":"2026-10-10","accessedAt":"2026-10-15","reuse":{"attribution":"Caiwei Chen, “How Chinese company DeepSeek released a top AI reasoning model despite US sanctions”, 2025-01-24.","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":"lg-src-gw-poolside-purpose-2023","title":"Poolside's journey to AGI","authors":["Jason Warner","Eiso Kant"],"publisher":"Poolside","publicationDate":"2023-10-07","version":"Dated page, retrieved on 2026-10-15 (SHA-256 b2e419615298aa7c2e6f00c5ff6e897bee7bffbd89911a94a24cfb06109fc5c6) directly from the publisher.","canonicalUrl":"https://poolside.ai/vision/purpose","evidenceUrl":"https://poolside.ai/vision/purpose","language":"en","archivedUrl":"https://web.archive.org/web/20261009091002/https://poolside.ai/vision/purpose","archivedAt":"2026-10-09","accessedAt":"2026-10-15","reuse":{"attribution":"Jason Warner, Eiso Kant, “Poolside's journey to AGI”, 2023-10-07.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"lg-src-new-cohere-about","title":"About Cohere (company page, undated)","authors":["Cohere"],"publisher":"Cohere","publicationDate":null,"version":"Undated page, retrieved on 2026-10-15 (SHA-256 cb2b697c4a6f6bf4ea036edbdd5b97bc6a55b02ec2e600ee264e21e1ed1cfa90) directly from the publisher. Its wording can change; earlier wording is not authenticated.","canonicalUrl":"https://cohere.com/about","evidenceUrl":"https://cohere.com/about","language":"en","archivedUrl":"https://web.archive.org/web/20261015184818/https://cohere.com/about","archivedAt":"2026-10-15","accessedAt":"2026-10-15","reuse":{"attribution":"Cohere, “About Cohere (company page, undated)”.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"lg-src-new-sakana-info","title":"Corporate Info — Sakana AI (company page, undated)","authors":["Sakana AI"],"publisher":"Sakana AI","publicationDate":null,"version":"Undated page, retrieved on 2026-10-15 (SHA-256 3ed2c29c0aaf2091548efeed35708539f04e7a56e97aa9169447772ee4156a3b) directly from the publisher. Its wording can change; earlier wording is not authenticated.","canonicalUrl":"https://sakana.ai/company-info/","evidenceUrl":"https://sakana.ai/company-info/","language":"en","archivedUrl":"https://web.archive.org/web/20261007203956/https://sakana.ai/company-info/","archivedAt":"2026-10-07","accessedAt":"2026-10-15","reuse":{"attribution":"Sakana AI, “Corporate Info — Sakana AI (company page, undated)”.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"source-cpo-ch-bis-uae-2026","title":"Enhanced Favorable Treatment for the United Arab Emirates Under the Export Administration Regulations (91 FR 43034)","authors":["Bureau of Industry and Security, U.S. Department of Commerce"],"publisher":"Bureau of Industry and Security, in the Federal Register (govinfo.gov)","publicationDate":"2026-07-14","version":"Retrieved for the candidate pack on 2026-10-28 (SHA-256 4e4b5555a41f7799f4e4dfe192a3a159c6fbb023da36ffacd76e4bbb80c5d9d5). Federal Register print of Tuesday, July 14, 2026, Vol. 91, No. 133, beginning at p. 43034, FR Doc. 2026-14132 (filed 7-10-26). Final rule. Retrieved again at admission on 2026-10-30 (SHA-256 4e4b5555a41f7799f4e4dfe192a3a159c6fbb023da36ffacd76e4bbb80c5d9d5, the same bytes).","canonicalUrl":"https://www.govinfo.gov/content/pkg/FR-2026-07-14/pdf/2026-14132.pdf","evidenceUrl":"https://www.govinfo.gov/content/pkg/FR-2026-07-14/pdf/2026-14132.pdf","language":"en","accessedAt":"2026-10-28","reuse":{"attribution":"Bureau of Industry and Security, U.S. Department of Commerce, “Enhanced Favorable Treatment for the United Arab Emirates Under the Export Administration Regulations (91 FR 43034)”, 2026-07-14.","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":"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-crusoe-microsoft-abilene-900mw","title":"Crusoe Announces New 900 MW AI Factory Campus in Abilene, Texas to Support Microsoft AI Infrastructure","authors":["Crusoe"],"publisher":"Crusoe","publicationDate":"2026-03-27","version":"Retrieved for the candidate pack on 2026-10-28 (SHA-256 9e03dfbc4734c549b95703fb0fd3b382cd090c00fb26182dc75197184da17ba0). Newsroom page as served on 2026-10-28; dateline \"ABILENE, TX — March 27, 2026\". Retrieved again at admission on 2026-10-30 (SHA-256 4a83ee6e049384ae45fa2fa9d31d2f624790da4bee2d76666e560d05b43abea0). Its wording can change; earlier wording is not authenticated.","canonicalUrl":"https://www.crusoe.ai/resources/newsroom/crusoe-announces-new-900-mw-ai-factory-campus-in-abilene-texas-to-support-microsoft-ai-infrastructure","evidenceUrl":"https://www.crusoe.ai/resources/newsroom/crusoe-announces-new-900-mw-ai-factory-campus-in-abilene-texas-to-support-microsoft-ai-infrastructure","language":"en","accessedAt":"2026-10-28","reuse":{"attribution":"Crusoe, “Crusoe Announces New 900 MW AI Factory Campus in Abilene, Texas to Support Microsoft AI Infrastructure”, 2026-03-27.","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-nscale-uk-2025","title":"Nscale announces UK AI infrastructure commitment in partnership with Microsoft, NVIDIA and OpenAI","authors":["Nscale"],"publisher":"Nscale","publicationDate":"2025-09-16","version":"Retrieved for the candidate pack on 2026-10-28 (SHA-256 0ed575206c677a25b438171a86990da09fd960f7cfcfedcc635a7ef8792a2333). Press release page as served on 2026-10-28; dateline \"LONDON, UK, Sep 16, 2025, 10:30 PM BST\". OpenAI's \"Introducing Stargate UK\" post refuses direct retrieval. Retrieved again at admission on 2026-10-30 (SHA-256 0ed575206c677a25b438171a86990da09fd960f7cfcfedcc635a7ef8792a2333, the same bytes). Its wording can change; earlier wording is not authenticated.","canonicalUrl":"https://www.nscale.com/press-releases/nscale-uk-ai-infrastructure-announcement","evidenceUrl":"https://www.nscale.com/press-releases/nscale-uk-ai-infrastructure-announcement","language":"en","accessedAt":"2026-10-28","reuse":{"attribution":"Nscale, “Nscale announces UK AI infrastructure commitment in partnership with Microsoft, NVIDIA and OpenAI”, 2025-09-16.","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-govuk-ne-aigz-2025","title":"North East England set for billions in investment and thousands of jobs as UK and US ink tech partnership","authors":["Department for Science, Innovation and Technology (GOV.UK)"],"publisher":"Department for Science, Innovation and Technology (GOV.UK)","publicationDate":"2025-09-16","version":"Retrieved for the candidate pack on 2026-10-28 (SHA-256 e3330e01113857a1ed5b3fa3b0c73abc8f75d4144b46e39f621b119fb3ce1597). GOV.UK press release as served on 2026-10-28; \"Published: 16 September 2025\", while the body says \"Announced today [17 September]\" (an embargoed release). The records date the designation by the body text. Retrieved again at admission on 2026-10-30 (SHA-256 31c1812a9fc4a58e15ba18e9c32a775347a9a37e5193ed5334c8f55d1b921e20).","canonicalUrl":"https://www.gov.uk/government/news/north-east-england-set-for-billions-in-investment-and-thousands-of-jobs-as-uk-and-us-ink-tech-partnership","evidenceUrl":"https://www.gov.uk/government/news/north-east-england-set-for-billions-in-investment-and-thousands-of-jobs-as-uk-and-us-ink-tech-partnership","language":"en","accessedAt":"2026-10-28","reuse":{"attribution":"Department for Science, Innovation and Technology (GOV.UK), “North East England set for billions in investment and thousands of jobs as UK and US ink tech partnership”, 2025-09-16.","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-supersedes-pf-2","sourceRole":"primary","support":"context","locator":"Opening paragraph, first sentence (opening words)","passage":"We’re releasing an update to our Preparedness Framework","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-03","note":"Verbatim excerpt checked by an automated research assistant on 2026-10-03 against the live page and the Internet Archive capture of 2025-04-16. The page names no earlier edition; the previous paraphrase (an update to a framework first released in 2023) overstated it and was replaced. Human editorial review is pending."},{"id":"evidence-supersedes-pf-2-changelog","sourceRole":"primary","support":"context","locator":"PDF p. 15 (printed p. 14), Appendix A “Change log”, opening sentence and item 2","passage":"The change log lists the updates made in this version, including removal of the low and medium levels, without naming the edition it updates.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-03","note":"Paraphrase checked by an automated research assistant on 2026-10-03 against the Internet Archive capture of 2025-04-15 and the current file; this page is identical in both. Context only: it implies an earlier edition but does not identify it. Human editorial review is pending."},{"id":"evidence-prov-rsp-v3-4-ceo-rso-decision-board-ltbt-informed","sourceRole":"primary","support":"supports","locator":"PDF p. 13 (printed p. 13), §3.4 Procedures, item 4 ('Governance notification')","passage":"Following approval of a Risk Report, the CEO and RSO will promptly share their decision(s), the underlying Risk Report, and internal feedback with both the Board and the LTBT.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-03","note":"Verbatim excerpt checked by an automated research assistant on 2026-10-03 against the inspected PDF, byte-identical to the Internet Archive capture of 2026-07-10. The text layer runs “RiskReport” together in the opening clause; the quote restores that space and otherwise matches the layer character for character. Extraction note: Full sentence quoted, as in the v3.0 and v3.2 records; wording identical to v3.0. Human editorial review is pending."},{"id":"evidence-cm-wh-2023-07-anthropic-1","sourceRole":"primary","support":"supports","locator":"First body paragraph after the two subheadings","passage":"Amazon, Anthropic, Google, Inflection, Meta, Microsoft, and OpenAI","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-04","note":"Verbatim quote, re-checked on 2026-10-04 by the integrating session against its own direct fetch of the cited expression (mechanical match after whitespace normalisation). Also matched in the Internet Archive capture of 2023-07-21. The passage was first proposed and machine-verified in the week 4b research pass. Human editorial review is pending."},{"id":"evidence-cm-wh-2023-07-anthropic-2","sourceRole":"primary","support":"context","locator":"Section “ASL-2 Commitments”, deployment measures, final paragraph (PDF p. 6)","passage":"The policy says its ASL-2 deployment measures overlap substantially with the White House voluntary commitments that Anthropic and other companies made in July, which it continues to maintain.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-04","note":"Editorial paraphrase, re-checked on 2026-10-04 by the integrating session against its own direct fetch of the cited expression. Added by the integrating session in week 4c; not part of the week 4b research pass. Human editorial review is pending."},{"id":"evidence-ts-b6-samuel-1960-contests","sourceRole":"secondary","support":"supports","locator":"NCBI efetch XML, Article/Abstract/AbstractText (two sentences)","passage":"The indexed abstract denies that the machine threatens humanity, says it has no will, and describes its conclusions as mechanistic consequences of its inputs.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-06","note":"Re-retrieved 2026-10-06; hash matched the pre-1996 candidate search. Recorded as a secondary source because the abstract’s authorship is not established."},{"id":"evidence-ts-b3-rlo-2019-intro","sourceRole":"primary","support":"supports","locator":"Paragraph previewing the fourth post","passage":"a sufficiently capable misaligned mesa-optimizer could learn to behave as if it were aligned without actually being robustly aligned. We will call this situation deceptive alignment.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-06","note":"Checked against the first stored version on 2026-10-06."},{"id":"evidence-ts-b2-cotra-2022","sourceRole":"primary","support":"supports","locator":"Summary bullet \"While humans are in control, Alex is incentivized to play the training game\"","passage":"The best way for Alex to maximize reward while under human control is to use its high situational awareness to deliberately appear safe and aligned at all times, while secretly manipulating humans in some cases.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-06","note":"The same passage as the B1 record."},{"id":"evidence-ec-m05-37","sourceRole":"primary","support":"supports","locator":"pdf page 9; 12.3 (printed p. 324)","passage":"The concern is not that robots will take human jobs and render humans unemployable. The traditional economic arguments against that are borne out by centuries of experience.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-14","note":"Candidate pack passage M05-37 (research/economics-mainstream). The quotation (27 words) matched the retrieved current copy (SHA-256 a43ecf48d5ec6131dfc002acab3d7df32dffb90bb9b7e41c30c470e14e6e68f6) in the pack's verification. Re-checked at admission on 2026-10-14 against a copy retrieved that day.","contextOnly":{"reason":"Fresh-context audit (A2) judged that the quotation misleads when shown alone.","requires":"The next sentences relocate the concern to prolonged 'short runs' with a large fraction not working, possibly for decades."}},{"id":"evidence-ec-e02-01","sourceRole":"primary","support":"supports","locator":"Hanson, \"Economic Growth Given Machine Intelligence\" (author-dated 1998; Wayback capture 2000-08-19), sec. 1 \"Introduction\", p. 1","passage":"Today, Artificial Intelligence and Robotics are long-standing fields of engineering whose explicit goals are to develop principles of design to eventually enable machines to accomplish all human cognitive and physical tasks.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage E02-01 (research/economics-mainstream/esoteric). The quotation (31 words) matched the Wayback capture 20000819034042 (SHA-256 754380e21b0e9f5a62fe71ea238a86e8640df48a0551c55cf4877a2a732b99c1) in the preparer's verification of 2026-10-11. It was not re-read in the correction pass of 2026-10-15 or at admission on 2026-10-16, because web.archive.org refused connections from the cloud environment. Admitted on the basis of the cross-family GPT check of 2026-10-15, which retrieved the same capture (Memento timestamp and SHA-256 equal to the recorded ones) and found the quotation matching."},{"id":"evidence-ec-e04-07","sourceRole":"primary","support":"supports","locator":"NBER WP 21547, sec. X.f, p. 39","passage":"We must therefore assume that those who develop the engines of superintelligence will eventually find they are sharing them with evil agents – with their military, commercial, and political adversaries.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-16","note":"Candidate pack passage E04-07 (research/economics-mainstream/esoteric). The quotation (30 words) matched the retrieved current copy (SHA-256 cb481e5e99ef9bad8575d9150bcc8464eda4904722cc5048389ab6159858b062) 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-gc-deepseek-founded-2023-07","sourceRole":"secondary","support":"supports","locator":"Paragraph beginning “Despite the buzz around R1”","passage":"Based in Hangzhou, China, it was founded in July 2023 by Liang Wenfeng","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The quotation (13 words) matched the retrieved text on 2026-10-15 and again in a fresh retrieval the same day. Read directly from the publisher. Matching ignored differences of whitespace, case, quotation marks, apostrophes, dashes and Unicode compatibility forms, so the source’s typography can differ. A cross-family machine check (2026-10-16/17) re-read the record’s sources and agreed with the record. No human editorial review.","retrievalRoute":"direct"},{"id":"lg-ev-new-ito-sakana-chairman","sourceRole":"primary","support":"supports","locator":"Section “Leadership”","passage":"Ren Ito Co-Founder & Chairman","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The quotation (5 words) matched the retrieved text on 2026-10-15 and again in a fresh retrieval the same day. Read directly from the publisher. Matching ignored differences of whitespace, case, quotation marks, apostrophes, dashes and Unicode compatibility forms, so the source’s typography can differ. No human editorial review.","retrievalRoute":"direct"},{"id":"lg-ev-gw-kant-cofounded-poolside","sourceRole":"primary","support":"supports","locator":"Byline","passage":"Eiso Kant CTO & Co-founder @ poolside","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-new-frosst-cohere-cofounder","sourceRole":"primary","support":"supports","locator":"Section “Meet our team”, Founders","passage":"Nick Frosst Co-founder","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The quotation (3 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":"ev-cmp-rsp-v3-3-d4-determination-1","sourceRole":"primary","support":"supports","locator":"PDF p. 13 (printed p. 12), §3.4 Procedures, item 3 ('Executive approval'), second sentence","passage":"CEO and RSO will make the ultimate determination regarding the adequacy of the risk assessment and any downstream deployment or development plans.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-27","note":"Verbatim excerpt checked by an automated research assistant on 2026-10-27 against the publisher copy, whose SHA-256 matches the copy retrieved on 2026-10-15. Human editorial review is pending."},{"id":"ev-cmp-rsp-v3-4-d6-competitor-1","sourceRole":"primary","support":"supports","locator":"PDF p. 17, Appendix A, table row 'Competitors have strong safety measures', Commitment column, first sentence","passage":"we will meet or exceed the overall risk reduction posture of these competitors","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-27","note":"Verbatim excerpt checked by an automated research assistant on 2026-10-27 against the publisher copy, whose SHA-256 matches the copy retrieved on 2026-10-15. Appendix A scenario 'Competitors have strong safety measures'. Human editorial review is pending."},{"id":"ev-cmp-rsp-v3-4-d6-competitor-2","sourceRole":"primary","support":"supports","locator":"PDF p. 17, Appendix A, table row 'General upleveling', Commitment column, second sentence","passage":"We will make a significant effort to meet or exceed that performance standard. However, we will not necessarily delay AI development and deployment in this scenario.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-27","note":"Verbatim excerpt checked by an automated research assistant on 2026-10-27 against the publisher copy, whose SHA-256 matches the copy retrieved on 2026-10-15. Appendix A scenario 'General upleveling': a competitor has implemented a significantly better risk mitigation that Anthropic could implement at comparable or lower cost. Human editorial review is pending."},{"id":"ev-cmp-meta-faf-v1-1-d4-determination-1","sourceRole":"primary","support":"supports","locator":"PDF p. 8 (printed p. 8), §2.1.3 Decide, 'Make a decision on release', second sentence","passage":"Informed by this analysis, a leadership team will either request further testing or information, require additional mitigations or improvements, or they will approve the model for release.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-27","note":"Passage recorded in the dataset for this edition, checked by an automated research assistant on 2026-10-03 against the Internet Archive capture of 2025-02-13, and matched against the same capture on 2026-10-27 by a reviewer from a different model family, which retrieved and hashed it (decoded PDF SHA-256 9fa301df…cc21; research/frontier-framework-comparison/codex-meta-v1-1-2026-10-27.md). Human editorial review is pending."},{"id":"ev-cmp-meta-faf-v1-1-d4-determination-2","sourceRole":"primary","support":"supports","locator":"PDF p. 13 (printed p. 13), §3.3 Risk thresholds, footnote 7 (attached to Table 1, High row, Risk Threshold column), third and final sentence","passage":"A final assessment of uplift is approved by senior-level decision-makers through our AI Risk process (see Section 2.1).","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-27","note":"Passage recorded in the dataset for this edition, checked by an automated research assistant on 2026-10-03 against the Internet Archive capture of 2025-02-13, and matched against the same capture on 2026-10-27 by a reviewer from a different model family, which retrieved and hashed it (decoded PDF SHA-256 9fa301df…cc21; research/frontier-framework-comparison/codex-meta-v1-1-2026-10-27.md). Human editorial review is pending."},{"id":"evidence-ap-erik-brynjolfsson-brynjolfsson-korinek-agrawal-2025-research-agenda-transformative-ai","sourceRole":"primary","support":"supports","locator":"Page head: author byline","passage":"Erik Brynjolfsson","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-27","note":"Credited name checked by an automated research assistant on 2026-10-27 in a direct retrieval. Human editorial review is pending.","retrievalRoute":"direct"},{"id":"evidence-ap-eliezer-yudkowsky-sl4-2006-02-15-yudkowsky-0602-14096","sourceRole":"primary","support":"supports","locator":"Message header, sender name (address not reproduced)","passage":"Eliezer S. Yudkowsky","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-27","note":"Credited name checked by an automated research assistant on 2026-10-27 in a direct retrieval and checked by a reviewer from a different model family on 2026-10-27. Human editorial review is pending.","retrievalRoute":"direct"},{"id":"evidence-ap-eliezer-yudkowsky-arbital-context-disaster","sourceRole":"primary","support":"supports","locator":"Author link in the post metadata beneath the title (Arbital mirror)","passage":"Eliezer Yudkowsky","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-27","note":"Passage and locator as inspected by a reviewer from a different model family on 2026-10-27 in a direct retrieval of the mirror page; the preparer’s retrieval of 2026-10-27 had failed, so the check’s reading replaces a restatement of the source record’s author list. The displayed credit does not establish sole authorship of every revision. Human editorial review is pending.","retrievalRoute":"direct"},{"id":"evidence-ap-eliezer-yudkowsky-lw-2018-challenges-to-christianos-capability-amplification-proposal","sourceRole":"primary","support":"supports","locator":"Post author list in the publisher’s post data (LessWrong GraphQL API)","passage":"Eliezer Yudkowsky","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-27","note":"Credited name checked by an automated research assistant on 2026-10-27 in a direct retrieval through the publisher’s post data and checked by a reviewer from a different model family on 2026-10-27. Human editorial review is pending.","retrievalRoute":"direct"},{"id":"ev-cmp-meta-aasf-v2-0-d2-levels-1","sourceRole":"primary","support":"supports","locator":"PDF p. 8 (printed p. 8), §2.2.1 Preparedness Reports, fourth paragraph, second sentence (opening)","passage":"The edition says Meta reports its risk assessment of the model, for plausible contexts of deployment, at one of three levels: “critical”, “high”, or “moderate or lower”.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-27","note":"Explicit paraphrase, written on 2026-10-27 by an automated research assistant after re-reading the passage in the publisher copy, whose SHA-256 matches the copy retrieved on 2026-10-15. The comparison quoted the passage; it is given here as a paraphrase to keep the edition within the quotation limit. Human editorial review is pending."},{"id":"evidence-cpo-ch-bis-uae-2026-diffusion-enforcement-1","sourceRole":"primary","support":"supports","locator":"Supplementary information, “Enforcement of License Requirements for Advanced Computing Items”, 91 FR 43036 (PDF p. 3)","passage":"Although § 742.6(a)(6)(iii)(A) of the EAR specifies a worldwide license requirement for certain advanced computing items (ECCNs 3A090.a, 4A090.a, and related ‘‘.z’’ paragraph items), consistent with guidance issued on May 13, 2025","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-30","note":"Candidate pack passage, compute, power and orbit (research/compute-power-orbit). The quotation (31 words) matched the text retrieved for the candidate pack on 2026-10-28 and again at admission on 2026-10-30, in a fresh retrieval whose SHA-256 equals the pack’s. Matching ignored differences of whitespace, case, quotation marks, dashes and Unicode compatibility forms, so the source’s typography can differ. A cross-family machine check (2026-10-29/30, item 71) re-read the record’s sources and found partial support; its corrections were applied at admission. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-cpo-ch-bis-uae-2026-diffusion-enforcement-2","sourceRole":"primary","support":"supports","locator":"same paragraph, 91 FR 43036","passage":"BIS is only enforcing this license requirement to destinations in Country Groups D:1, D:4, or D:5","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-30","note":"Candidate pack passage, compute, power and orbit (research/compute-power-orbit). The quotation (16 words) matched the text retrieved for the candidate pack on 2026-10-28 and again at admission on 2026-10-30, in a fresh retrieval whose SHA-256 equals the pack’s. Matching ignored differences of whitespace, case, quotation marks, dashes and Unicode compatibility forms, so the source’s typography can differ. A cross-family machine check (2026-10-29/30, item 71) re-read the record’s sources and found partial support; its corrections were applied at admission. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-cpo-ch-bis-uae-2026-diffusion-enforcement-3","sourceRole":"primary","support":"supports","locator":"same paragraph, 91 FR 43036 (PDF p. 3)","passage":"(excluding those also specified in Country Groups A:5 or A:6), and to entities headquartered in or with an ultimate parent headquartered in Country Group D:5 or Macau, wherever located","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-30","note":"Added at the admission of the compute, power and orbit pack (research/compute-power-orbit). The exclusions and the ultimate-parent condition of the same sentence (cross-family check, item 71). The quotation (29 words) matched a direct retrieval of 2026-10-30 (SHA-256 4e4b5555a41f7799f4e4dfe192a3a159c6fbb023da36ffacd76e4bbb80c5d9d5). Matching ignored differences of whitespace, case, quotation marks, dashes and Unicode compatibility forms. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-cpo-ch-bis-uae-2026-diffusion-enforcement-4","sourceRole":"primary","support":"supports","locator":"same paragraph, 91 FR 43036 (PDF p. 3)","passage":"Nevertheless, BIS will continue enforcing the license requirement for exports, reexports, and transfers (incountry) of advanced computing items specified in § 742.6(a)(6)(iii)(A) to or within the UAE, except when the ultimate consignee and all end users are UAE Government entities or approved commercial entities","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-30","note":"Added at the admission of the compute, power and orbit pack (research/compute-power-orbit). The retained UAE licence requirement and its exceptions, stated next in the same paragraph (cross-family check, item 71). The text layer prints “incountry” for “in-country”. The quotation (43 words) matched a direct retrieval of 2026-10-30 (SHA-256 4e4b5555a41f7799f4e4dfe192a3a159c6fbb023da36ffacd76e4bbb80c5d9d5). Matching ignored differences of whitespace, case, quotation marks, dashes and Unicode compatibility forms. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-cpo-dc-crusoe-abilene-total-2-1gw-1","sourceRole":"primary","support":"supports","locator":"paragraph 1","passage":"bringing the total projected capacity across the full Abilene site to 2.1 gigawatts (GW)","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 4a83ee6e049384ae45fa2fa9d31d2f624790da4bee2d76666e560d05b43abea0; 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-uk-openai-offtake-1","sourceRole":"primary","support":"supports","locator":"paragraph 4","passage":"OpenAI will explore offtake of up to 8,000 NVIDIA GPUs in Q1 2026 with the potential to scale to 31,000 NVIDIA GPUs over time","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-30","note":"Candidate pack passage, compute, power and orbit (research/compute-power-orbit). The quotation (24 words) matched the text retrieved for the candidate pack on 2026-10-28 and again at admission on 2026-10-30, in a fresh retrieval whose SHA-256 equals the pack’s. Matching ignored differences of whitespace, case, quotation marks, dashes and Unicode compatibility forms, so the source’s typography can differ. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-cpo-dc-stargate-uk-openai-offtake-2","sourceRole":"primary","support":"supports","locator":"paragraph 9","passage":"The first phase will see OpenAI take up to 8,000 GPUs","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-30","note":"Candidate pack passage, compute, power and orbit (research/compute-power-orbit). The quotation (11 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 31c1812a9fc4a58e15ba18e9c32a775347a9a37e5193ed5334c8f55d1b921e20; 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. Stronger stage wording than the Nscale release; recorded as a difference of wording. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-cpo-dc-stargate-oracle-sites-capacity-1","sourceRole":"primary","support":"supports","locator":"paragraph 2","passage":"can deliver over 5.5 gigawatts of capacity","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-30","note":"Candidate pack passage, compute, power and orbit (research/compute-power-orbit). The quotation (7 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. A cross-family machine check (2026-10-29/30, item 22) 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-oracle-sites-capacity-2","sourceRole":"primary","support":"context","locator":"paragraph 2","passage":"the aggregate covers the three new sites \"combined with an additional potential expansion of 600 megawatts near the flagship Stargate site in Abilene, Texas\".","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. A cross-family machine check (2026-10-29/30, item 22) answered the record’s question blind from fresh retrievals and agreed with the record. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-ab-fw-meta-faf-v1-1-cb-risk-process-1","sourceRole":"primary","support":"supports","locator":"PDF pp. 3, 8, 13 and 16–18; §2.1.2 and Table 1","passage":"Chemical and biological risks are in scope; red teaming is conducted when relevant capability levels are reached, with external experts where appropriate; and release decisions rest on residual risk after evaluations and mitigations, with leadership able to require further testing, information or mitigation.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-30","note":"Added at the admission of the AI and biosecurity pack (research/ai-bio). The edition could not be read for the candidate pack or at admission: the publisher’s address failed and the Internet Archive refused this environment throughout. On 2026-10-30 the cross-family machine check opened the Internet Archive capture of 13 February 2025 (web.archive.org/web/20250213105620/https://ai.meta.com/static-resource/meta-frontier-ai-framework/), which is the expression the dataset source already names, and reported the edition’s policy-level provisions. This record rests on that reading, which this admission could not repeat. The passage is Selekoa’s policy-level paraphrase of what the check reported, not a quotation of the edition. No human editorial review."},{"id":"evidence-ab-ev-chatgpt-agent-card-published-1","sourceRole":"primary","support":"supports","locator":"PDF p. 1, cover","passage":"ChatGPT Agent System Card OpenAI July 17, 2025","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-30","note":"Candidate pack passage, AI and biosecurity (research/ai-bio). The quotation (8 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. Cover lines, run together in the PDF text layer. A cross-family machine check (2026-10-30, item 19) answered the record’s question blind from fresh retrievals and agreed with the record. No human editorial review.","retrievalRoute":"direct"}],"at":{"entities":[0,1,19,28,48,49,53,54,59,163,172,199,200,250,263,324,330,335,343,346,424,485,517,524,526,552,586,588,612,664,752,770,776,785,859,1090,1142,1181,1244],"claims":[85,427,674,786,962,978,1386,1503,1536,1653,1923,1988,2016,2086,2094,2151,2199,2427,2454,2459,2700,3072,3121,3148,3154,3358,3387],"sources":[33,35,42,43,49,178,180,190,237,263,329,335,340,391,452,475,489,505,562,618,683,723,729,914,961,963,968,989],"evidence":[101,102,459,850,851,996,1185,1202,1730,1876,1919,2083,2486,2567,2605,2698,2711,2712,2793,2794,2844,3074,3101,3106,3393,4030,4031,4032,4033,4121,4181,4182,4195,4196,4608,4652]},"version":"1lg47t7"}