{"entities":[{"id":"anthropic","name":"Anthropic"},{"id":"google-deepmind","name":"Google DeepMind"},{"id":"xai","name":"xAI"},{"id":"deepmind","name":"DeepMind"},{"id":"google-brain","name":"Google Brain"},{"id":"eliezer-yudkowsky","name":"Eliezer Yudkowsky"},{"id":"rsp-v2-0","name":"Anthropic Responsible Scaling Policy — v2.0"},{"id":"rsp-v2-2","name":"Anthropic Responsible Scaling Policy — v2.2"},{"id":"rsp-v3-2","name":"Anthropic Responsible Scaling Policy — v3.2"},{"id":"fsf-v3-1","name":"Google DeepMind Frontier Safety Framework — v3.1"},{"id":"frontier-governance-2026","name":"OpenAI Frontier Governance Framework — May 2026 edition"},{"id":"faif-2025","name":"xAI Frontier Artificial Intelligence Framework — December 2025 edition"},{"id":"faif-2026","name":"xAI Frontier Artificial Intelligence Framework — June 2026 edition"},{"id":"thinking-machines-lab","name":"Thinking Machines Lab"},{"id":"mira-murati","name":"Mira Murati"},{"id":"christian-szegedy","name":"Christian Szegedy"},{"id":"california-sb-53","name":"California SB 53 (2025): Transparency in Frontier Artificial Intelligence Act, as chaptered"},{"id":"anthropic-ltbt","name":"Anthropic Long-Term Benefit Trust"},{"id":"minimax","name":"MiniMax"},{"id":"vinge-1993-coming-technological-singularity","name":"The Coming Technological Singularity: How to Survive in the Post-Human Era"},{"id":"staring-into-the-singularity","name":"Staring into the Singularity"},{"id":"christiano-2019-what-failure-looks-like","name":"What failure looks like"},{"id":"observed-behaviour-as-evidence","name":"Observed behaviour under training or oversight as evidence"},{"id":"competition-defeats-restraint","name":"Competition defeats restraint"},{"id":"risk-is-film-fiction","name":"AI risk as film science fiction"},{"id":"human-data-humanlike-mind","name":"Trained on human data, so humanlike"},{"id":"extropians-1996-11-13-yudkowsky-2941","name":"Re: Zones of Thought (was Zindell's _The Wild_)"},{"id":"lw-2025-re-recent-anthropic-safety-research","name":"Re: recent Anthropic safety research"},{"id":"compensating-policy-for-ai","name":"Compensating policy for AI’s distributional effects"},{"id":"acemoglu-restrepo-2018-race-between-man-and-machine","name":"The Race between Man and Machine: Implications of Technology for Growth, Factor Shares, and Employment"},{"id":"reuther-1955-automation-hearings-statement","name":"Statement (president, Congress of Industrial Organizations), in Automation and Technological Change: Hearings"},{"id":"ai-developer-incentives-and-safety","name":"AI developers’ incentives matter for safety"},{"id":"hanson-2008-emulations-go-foom","name":"Emulations Go Foom"},{"id":"aidan-gomez","name":"Aidan Gomez"},{"id":"gemini-app-team","name":"Gemini app team"},{"id":"kanika-bahl","name":"Kanika Bahl"},{"id":"mike-krieger","name":"Mike Krieger"},{"id":"moonshot-ai","name":"Moonshot AI / 月之暗面"},{"id":"yang-zhilin","name":"Yang Zhilin"},{"id":"zach-robinson","name":"Zach Robinson"},{"id":"daron-acemoglu","name":"Daron Acemoglu"},{"id":"aiia-ai-safety-commitments-2024","name":"Artificial Intelligence Safety Commitments (人工智能安全承诺), December 2024"},{"id":"hongshan-capital","name":"HongShan (HSG)"},{"id":"us-nsm-ai-2024","name":"Memorandum on Advancing the United States' Leadership in Artificial Intelligence; Harnessing Artificial Intelligence to Fulfill National Security Objectives; and Fostering the Safety, Security, and Trustworthiness of Artificial Intelligence"},{"id":"grok-4-model-card","name":"Grok 4 Model Card"}],"claims":[{"id":"prov-rsp-v3-2-ltbt-regular-briefing","statement":"RSP v3.2 states that Anthropic will regularly brief the LTBT on plans and developments related to its Risk Reports, including model training, capability evaluations, mitigations and risk analyses.","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-v3-2-ltbt-regular-briefing"],"attributes":{"provision":{"kind":"decision_authority","key":"rsp-ltbt-regular-briefing","parties":[{"role":"informed","actor":"Long-Term Benefit Trust"}]}}},{"id":"prov-fsf-v3-1-response-plan","statement":"FSF v3.1 states that when a model reaches an alert threshold for a CCL, Google DeepMind will assess the model's proximity to the CCL and analyse the risk posed, involving internal and external experts as needed, and that this will inform the formulation and application of a response plan.","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-v3-1-response-plan"],"attributes":{"provision":{"kind":"required_action","key":"fsf-response-plan"}}},{"id":"sb-transparency-report","statement":"SB 53 requires a frontier developer, before or at the same time as deploying a new or substantially modified frontier model, to publish clearly and conspicuously on its website a transparency report with listed basic information. A large frontier developer must also include summaries of its catastrophic-risk assessments under its frontier AI framework, their results, the extent of third-party evaluator involvement, and other steps taken to fulfil its framework for that model.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"What the enacted text states, quoted from the chaptered expression. It is not evidence that the statute applies to any named developer, or of compliance, enforcement or effect. 'Deploy' means making a frontier model available to a third party for use, modification, copying or combination with other software. It excludes making the model available to a third party for the primary purpose of developing or evaluating it (§ 22757.11(e)). The statute does not define 'substantially modified'. A large frontier developer's framework must describe how it decides when its models are substantially modified enough to require these disclosures (§ 22757.12(a)(6)). Publishing the information as part of a larger document, including a system card or model card, is deemed compliance with the applicable paragraph (§ 22757.12(c)(3)). A frontier developer is encouraged, but not required, to make these disclosures consistent with, or superior to, industry best practices (§ 22757.12(c)(4)). Published documents may be redacted where necessary to protect trade secrets, cybersecurity, public safety or US national security, or to comply with federal or state law. The developer must describe the character and justification of each redaction, to the extent those concerns permit, and keep the unredacted information for five years (§ 22757.12(f)). A statutory requirement is not evidence that any report was published, or that any report was complete or adequate.","evidenceIds":["evidence-sb-transparency-report-1","evidence-sb-transparency-report-2"],"attributes":{"provision":{"kind":"required_action","key":"sb53-transparency-report-before-deployment"}}},{"id":"sb-system-card-deemed-compliance","statement":"SB 53 provides that a frontier developer that publishes the required transparency-report information as part of a larger document, such as a system card or model card, is deemed in compliance with the relevant paragraph.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"What the enacted text states, quoted from the chaptered expression. It is not evidence that the statute applies to any named developer, or of compliance, enforcement or effect. This record states the statutory rule only. Whether a particular system card contains the required information is a compliance question and is not recorded.","evidenceIds":["evidence-sb-system-card-deemed-compliance-1"],"attributes":{"provision":{"kind":"exception","key":"sb53-transparency-report-larger-document"}}},{"id":"ts-c3-vinge-1993","statement":"Vinge’s 1993 paper argues that the competitive advantage of every advance in automation is so compelling that laws or customs forbidding it merely ensure that someone else gets it first.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"A conditional argument; the fictional prohibition it mentions is an illustration, not a real policy.","evidenceIds":["evidence-ts-c3-vinge-1993"],"attributes":{"thought":{"roles":["earliest-located"],"register":"nonfiction","claimed":{"date":"1993-03-30","note":"Presented at the VISION-21 Symposium, 30–31 March 1993. The delivered version was not retrieved, so the proceedings wording is not dated to the presentation."}}}},{"id":"ts-b5-christiano-2019-wflf-context","statement":"Christiano’s post of 2019-03-17 says influence-seeking policies would also score well on the training objective.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The same passage as the B3 record.","evidenceIds":["evidence-ts-b5-christiano-2019-wflf-context"],"attributes":{"thought":{"roles":["context"],"register":"nonfiction","missing":"It says influence-seeking policies would also score well in training, but draws no explicit conclusion about what good scores show."}}},{"id":"yk-y20-staring-into-the-singularity-b","statement":"“Staring into the Singularity” (1996-11-18) bears on Y20. Rejects the science-fiction picture of AIs acquiring resentment; in this early text the point supports the conclusion that hostile Powers are not worth worrying about, while a badly programmed Power with hostile goals is called 'an engineering risk'.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Premise addressed: The concern is about robots that come to hate or rebel against humans. Wayback capture 2000-12-04 (page dated 11/18/1996, updated 08/31/2000); first appearance of the sentence not established. Related rather than answered-it because the surrounding argument draws a reassuring conclusion that his later texts do not share. In context the point reassures: the same paragraph says there is no point worrying that hostile Powers will inevitably wipe out humanity. Not the later \"risk is real but not Terminator\" argument.","evidenceIds":["evidence-yk-t20-c"],"attributes":{"thought":{"roles":["context"],"register":"nonfiction","missing":"It rejects hostile emotions arising in AI in order to reassure; it does not argue that the risk is film fiction or that films are a fair guide.","attested":{"date":"2000-12-04","note":"The quoted wording is that of version 1.2.1 (\"Updated: 08/31/2000\") as captured on 2000-12-04; no capture of the 1996–1999 page holds the text."}}}},{"id":"yk-y21-lw-2025-re-recent-anthropic-safety-research-c","statement":"The post “Re: recent Anthropic safety research” (2025-08-06) contests Y21. Commenting on reported behaviour of a language model, he distinguishes a learned human-like persona from the system's underlying preferences, which he expects to be alien.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Premise addressed: Training on human text or behaviour yields human-like thought and motives. Crossposted from X by the LessWrong team on 2025-08-06; the linked X status ID decodes to 2025-08-04 (the X original was not retrieved), and the posting-day version is unavailable from the API. Hedged: the next paragraph begins \"I could be wrong.\"","evidenceIds":["evidence-yk-t21-e"],"attributes":{"thought":{"roles":["contesting","later-restatement"],"register":"nonfiction","claimed":{"date":"2025-08-04","note":"Decoded from the linked X status identifier; the X original was not retrieved and the date is unverified (cross-family check, 2026-10-07)."},"engagement":{"kind":"contrary-only","rationale":"Under the message-level rule, the passage states the contrary conclusion without giving a reason against the premise that training on human text or behaviour yields human-like thought and motives, even when read with its message or dialogue turn."}}}},{"id":"ec-m06b-reuther-1955-automation-hearings-statement-a","statement":"Reuther, “Statement (president, Congress of Industrial Organizations), in Automation and Technological Change: Hearings” (1955-10-17) articulates M06b. Reuther urges the subcommittee to consider earlier Social Security payments for workers forced into retirement before 65 because technological changes have taken their jobs and their age keeps them from finding other work.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Added on 2026-10-25 after the same-family and cross-family re-checks of the economics admission (maintainer decision of 2026-10-25). A proposal to the subcommittee in his prepared statement, not enacted policy or evidence of implementation.","evidenceIds":["evidence-ec-m06b-reuther-a"],"attributes":{"thought":{"roles":["articulation"],"register":"nonfiction","tier":{"value":2,"basis":"Witness who is not an economist: Walter P. Reuther, president of the Congress of Industrial Organizations."}}}},{"id":"ec-e04-hanson-2008-emulations-go-foom-b","statement":"Hanson, “Emulations Go Foom” (2008-11-22), in “The Hanson-Yudkowsky AI-Foom Debate”, bears on E04. Hanson's scenario of competing whole-brain-emulation (\"bot\") projects.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Hanson's scenario about emulations, not designed AI. The race argument bears on whether one developer could dominate, not directly on safety. Tier 2.","evidenceIds":["evidence-ec-e04-02"],"attributes":{"thought":{"roles":["context"],"register":"nonfiction","missing":"The race is analysed for leads and dominance; the safety or control of the systems is not discussed here.","tier":{"value":2,"basis":"Blog debate (2008 Overcoming Bias posts) between an economist and a non-economist: tier 2 as an online debate; the maintainer confirmed tier 2 for the debate on 2026-10-12."}}}},{"id":"lg-goo-deepmind-founding-stated-2010","statement":"Google DeepMind's organisation page states that DeepMind started in 2010; Demis Hassabis's message of 20 April 2023 also dates DeepMind's launch to 2010.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The sources state a start year only; they do not establish a legal incorporation date or the founder roster, which existing records cover.","evidenceIds":["lg-ev-goo-deepmind-founding-about","lg-ev-goo-deepmind-founding-hassabis-2023"]},{"id":"lg-new-gomez-brain-2017","statement":"Attention Is All You Need (arXiv v1, 12 June 2017) lists Aidan N. Gomez with a University of Toronto affiliation and a footnote that his work was performed while at Google Brain.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The footnote marker (†) on Gomez’s name links him to the note; the paper does not give a title or dates for his time at Google Brain.","evidenceIds":["lg-ev-new-gomez-brain-2017","lg-ev-new-gomez-brain-2017-b"],"attributes":{"role":"Paper author; work performed while at Google Brain"}},{"id":"lg-ant-bahl-ltbt-trustee","statement":"Anthropic's announcement of its Long-Term Benefit Trust on 19 September 2023 names Kanika Bahl, CEO and President of Evidence Action, as an initial Trustee.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"A documented role at the announcement date. The page says the board chose the initial Trustees and that Trustees serve one-year terms; reappointments are not documented here.","evidenceIds":["lg-ev-ant-bahl-ltbt-trustee-1"],"attributes":{"role":"Trustee (initial Trustee)"}},{"id":"lg-cn-yang-zhilin-moonshot-ceo-2024","statement":"The South China Morning Post identified Yang Zhilin on 7 August 2024 as Moonshot AI’s chief executive.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The record rests on reporting; a dated identification, not a company announcement, and not evidence of a current role.","evidenceIds":["lg-ev-cn-yang-zhilin-moonshot-ceo-2024"],"attributes":{"role":"chief executive"}},{"id":"lg-goo-gemini-app-team-to-gdm-2024","statement":"Sundar Pichai announced on 17 October 2024 that the Gemini app team would join Google DeepMind under Demis Hassabis.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"An announced move; the note names the team's lead only by first name (Sissie), and the record does not establish completion.","evidenceIds":["lg-ev-goo-gemini-app-team-to-gdm-2024"],"attributes":{"activityStage":"announced"}},{"id":"lg-mx-szegedy-xai-departure-2025","statement":"TechCrunch reported on 11 February 2026 that Christian Szegedy had left xAI in February 2025.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The record rests on retrospective reporting; no post by Szegedy or statement by xAI was retrieved. It does not establish his next position.","evidenceIds":["lg-ev-mx-szegedy-xai-departure-2025","lg-ev-mx-szegedy-xai-departure-2025-team"],"attributes":{"role":"Founding-team member"}},{"id":"lg-ant-robinson-ltbt-departure-2026","statement":"Anthropic's Long-Term Benefit Trust page states, in an undated footnote, that Zach Robinson stepped down from the Trust in January 2026; Anthropic stated that he did so to focus on non-profit and philanthropic work.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The footnote is undated and was added to the page after its original publication; it gives a month, not a day. The stated reason is Anthropic's wording.","evidenceIds":["lg-ev-ant-robinson-ltbt-departure-2026","lg-ev-ant-robinson-ltbt-departure-2026-initial"],"attributes":{"role":"Trustee (initial Trustee)","statedReason":{"by":"anthropic","text":"Anthropic's page states that he stepped down to focus on non-profit and philanthropic work.","evidenceId":"lg-ev-ant-robinson-ltbt-departure-2026"}}},{"id":"lg-ant-krieger-labs-2026","statement":"Anthropic announced on 13 January 2026 that Mike Krieger was joining Labs to build alongside Ben Mann.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The announcement gives no title for his Labs role and does not state when his CPO role ended. Ben Mann's own role in Labs is not recorded from this sentence.","evidenceIds":["lg-ev-ant-krieger-labs-2026"],"attributes":{"activityStage":"announced","role":"Member of Labs"}},{"id":"lg-new-murati-tml-cofounder","statement":"Thinking Machines Lab’s announcement of 10 March 2026 identifies Mira Murati as its cofounder.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"A company description of its own founder; it does not establish the founding date or the full founder roster.","evidenceIds":["lg-ev-new-murati-tml-cofounder"]},{"id":"cmp-faif-2025-d3-halt","statement":"FAIF · December 2025 edition states a provision on the comparison question “Pause, non-deployment or non-training”, 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. This is an incident-response option ('may') in the event of an imminent threat, not a stated consequence of reaching a threshold. The deployment acceptance criteria (“Named levels or thresholds”) are not accompanied by an explicit statement that deployment will not proceed if they are not met. 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-faif-2025-d3-halt-1"],"attributes":{"provision":{"kind":"required_action","key":"comparison-d3-halt","facet":{"id":"D3.halt","status":"partly"}}}},{"id":"cmp-faif-2026-d4-determination","statement":"No passage meeting the comparison question “Determination of capability or safeguard adequacy” 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 person, office or body the edition names as making or approving the determination that a threshold is reached, that safeguards are adequate, or that a model may be deployed. 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 decid/decision, approv, risk owner, leadership, officer, determin. Hits examined and not counted: PDF p. 6 'The final decision also reflects an overall judgment based on all the available evidence' does not name who decides (recorded under “Other exceptions or discretion”); PDF pp. 2, 5, 6 'determine/determination' refer to xAI generically; PDF p. 8 §3 designates risk owners for mitigating risks and monitoring incidents, not for threshold, safeguard or deployment determinations; PDF p. 9 §4 refers to 'internal governance structures' without naming a body. 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. 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-d4-determination"],"attributes":{"provision":{"kind":"decision_authority","key":"comparison-d4-determination","facet":{"id":"D4.determination"}}}},{"id":"ap-daron-acemoglu-acemoglu-restrepo-2018-race-between-man-and-machine","statement":"“The Race between Man and Machine: Implications of Technology for Growth, Factor Shares, and Employment” (2018-06) credits Daron Acemoglu as one of its 2 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-daron-acemoglu-acemoglu-restrepo-2018-race-between-man-and-machine"]},{"id":"ap-eliezer-yudkowsky-extropians-1996-11-13-yudkowsky-2941","statement":"The mailing-list message “Re: Zones of Thought (was Zindell's _The Wild_)” (1996-11-13) is sent under the name Eliezer 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-extropians-1996-11-13-yudkowsky-2941"]},{"id":"cmp-rsp-v2-0-d8-changes","statement":"RSP · v2.0 states a provision on the comparison question “Stated changes from the previous edition”.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"What the edition states, paraphrased from the cited expression with its page. It is not evidence of implementation, compliance or effectiveness, and a shared question does not make two editions’ provisions equivalent.","evidenceIds":["ev-cmp-rsp-v2-0-d8-changes-1"],"attributes":{"provision":{"kind":"revision","key":"comparison-d8-changes","facet":{"id":"D8.changes","status":"stated"}}}},{"id":"cmp-rsp-v2-2-d2-levels","statement":"RSP · v2.2 states a provision on the comparison question “Named levels or thresholds”.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"Levels are split into labelled Capability Thresholds (capabilities) and ASL Standards (safeguards); §1 (PDF p. 7) describes a Capability Threshold as the trigger for moving up from ASL-N to the ASL-N+1 Standard or higher. ASL-2 and ASL-3 Standards are specified; ASL-4 Standards are referenced as expected requirements to be detailed later. What the edition states, paraphrased from the cited expression with its page. It is not evidence of implementation, compliance or effectiveness, and a shared question does not make two editions’ provisions equivalent.","evidenceIds":["ev-cmp-rsp-v2-2-d2-levels-1","ev-cmp-rsp-v2-2-d2-levels-2"],"attributes":{"provision":{"kind":"threshold","key":"comparison-d2-levels","facet":{"id":"D2.levels","status":"stated"}}}},{"id":"cmp-frontier-governance-2026-d1-autonomy","statement":"FGF · May 2026 edition states a provision on the comparison question “Autonomy, self-improvement or AI research and development”.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The domain is framed as loss of control, with a three-tier table; self-improvement is named within it rather than as its own category. What the edition states, paraphrased from the cited expression with its page. It is not evidence of implementation, compliance or effectiveness, and a shared question does not make two editions’ provisions equivalent.","evidenceIds":["ev-cmp-frontier-governance-2026-d1-autonomy-1","ev-cmp-frontier-governance-2026-d1-autonomy-2"],"attributes":{"provision":{"kind":"risk_domain","key":"comparison-d1-autonomy","facet":{"id":"D1.autonomy","status":"stated"}}}},{"id":"c2-governance-aiia-aisc-signatory-minimax","statement":"MiniMax is listed, as 上海稀宇科技有限公司, among the 17 companies on the initial signatory list (首批签署企业名单) of the AIIA AI Safety Commitments posted by CAICT on 24 December 2024.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The list names the legal entity 上海稀宇科技有限公司. The record maps it to the group-level entity MiniMax, which carries that name as an alias; the mapping rests on the company name, and no claim is made about group structure or about which group company is bound. A listing is the publisher’s statement, dated by the post; it is not a signature date and not evidence that the company complied with or implemented the commitments. A later total of 22 signatories is reported, but no later list was retrieved.","evidenceIds":["c2-ev-governance-aiia-aisc-signatory-minimax-1","c2-ev-governance-aiia-aisc-signatory-minimax-2"],"attributes":{"commitment":{"stage":"listed"}}},{"id":"c2-minimax-investor-hongshan","statement":"MiniMax’s prospectus lists HSG Growth VII Holdco E, Ltd. and Himalia Holding Limited, whose sole shareholders it names as HongShan Capital Growth Funds VII and VI, among its other key Pre-IPO Investors.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"The prospectus does not, in the cited passage, attribute any round amount to this investor; no round total is attributed to it. Shareholding percentages are those the filing states for specified dates. No influence or control is inferred from the shareholding.","evidenceIds":["c2-ev-minimax-investor-hongshan"],"attributes":{"transactionType":"equity_investment","relationshipContext":"external","period":"Pre-IPO Investments (series not specified in the cited passage)","transactionStage":"invested"}},{"id":"ab-gov-us-nsm-aisi-two-models","statement":"Subsection 3.3(e)(i) of the national security memorandum of 24 October 2024 directs the AI Safety Institute, within 180 days and subject to private sector cooperation, to pursue voluntary preliminary testing of at least two frontier AI models before their public deployment or release, and provides that the testing shall assess models' capabilities to aid offensive cyber operations, accelerate development of biological and/or chemical weapons, autonomously carry out malicious behaviour, automate development and deployment of other models with such capabilities, and give rise to other risks identified by the institute.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"A directive conditioned on private sector cooperation. It names no model and no developer, and it is not evidence that any testing was performed; evaluations actually published are recorded elsewhere in this pack's evaluations strand. The memorandum's list of capabilities to be assessed ends with risks identified by the institute, so it is not a closed list. A presidential memorandum of 5 June 2026 states that it rescinds and replaces this memorandum and its associated guidance (recorded separately). The provision stays on record as the 2024 text stated it, and no implementation, continuing force or wider legal consequence is inferred.","evidenceIds":["evidence-ab-gov-us-nsm-aisi-two-models-1","evidence-ab-gov-us-nsm-aisi-two-models-2"],"attributes":{"provision":{"kind":"evaluation_timing","key":"us-nsm-ai-2024-sec33ei-preliminary-testing"}}},{"id":"ab-ev-grok4-biology-result","statement":"The Grok 4 Model Card states that the area of highest concern is the model's expert-level biology capabilities, which significantly exceed human expert baselines on the benchmarks used, while stating that these evaluations measure dual-use knowledge and that a high score indicates greater capability to enable weapons development, not necessarily increased risk.","derivation":"explicit","resolution":"supported","editorialState":"seed","qualification":"A benchmark result and its reading as the developer states them, for the model versions and public benchmarks the card names. It is not a finding about real-world capability or risk, and the card itself says the scores measure dual-use knowledge rather than risk. Safeguards were removed for these measurements, as the card states. Benchmark names and scores are not reproduced beyond the card's own summary sentences; no task, protocol or agent is recorded.","evidenceIds":["evidence-ab-ev-grok4-biology-result-1","evidence-ab-ev-grok4-biology-result-2","evidence-ab-ev-grok4-biology-result-3"],"attributes":{"evaluatedSystem":"Grok 4 API and Grok 4 Web, with safeguards removed for the dual-use capability assessment","evaluation":{"what":"Dual-use chemical and biological knowledge, measured on public benchmarks","how":"Benchmark scoring against stated human expert baselines, with safeguards removed for the assessment","result":"The card states that biology capabilities are the area of highest concern and significantly exceed human expert baselines, and that chemistry capabilities are strong","caveats":"The card states that these evaluations measure dual-use knowledge, so a high score indicates greater capability to enable weapons development and not necessarily increased risk; that radiological and nuclear capabilities are not evaluated; and that, given existing nonproliferation regimes, it assesses its models as generally posing a low risk of enabling malicious use."}}}],"sources":[{"id":"source-deepmind-combination","title":"Announcing Google DeepMind","authors":["Demis Hassabis"],"publisher":"Google DeepMind","publicationDate":"2023-04-20","version":"Dated page, inspected at retrieval","canonicalUrl":"https://deepmind.google/blog/announcing-google-deepmind/","evidenceUrl":"https://deepmind.google/blog/announcing-google-deepmind/","language":"en","archivedUrl":"https://web.archive.org/web/20260904134502/https://deepmind.google/blog/announcing-google-deepmind/","archivedAt":"2026-09-04","accessedAt":"2026-09-17","reuse":{"attribution":"Google DeepMind, Announcing Google DeepMind (2023-04-20).","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and minimal attributed excerpts only. Recheck terms before wider redistribution."}},{"id":"source-rsp-2-0","title":"Responsible Scaling Policy — v2.0","authors":["Anthropic"],"publisher":"Anthropic","publicationDate":null,"version":"Responsible Scaling Policy v2.0; effective date 2024-10-15. v2.0 identity is supplied by the official archive and later dated changelogs; the cover is not numbered.","versionDate":"2024-10-15","canonicalUrl":"https://www-cdn.anthropic.com/616dee633636e5bd309cb73aed8622e80fe47839.pdf","evidenceUrl":"https://www-cdn.anthropic.com/616dee633636e5bd309cb73aed8622e80fe47839.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20241022154907/https://assets.anthropic.com/m/24a47b00f10301cd/original/Anthropic-Responsible-Scaling-Policy-2024-10-15.pdf","archivedAt":"2024-10-22","accessedAt":"2026-09-18","reuse":{"attribution":"Anthropic, Responsible Scaling Policy — v2.0.","licence":"No open redistribution licence asserted","note":"Selekoa practice: metadata, attributed paraphrases and short attributed quotations under the editorial policy (at most 50 words each; at most 400 quoted words per document across the dataset, and about 10% of a document under about 2,000 words). Documents are linked externally, not redistributed. Publisher terms that forbid quotation override this practice."}},{"id":"source-rsp-2-2","title":"Responsible Scaling Policy — v2.2","authors":["Anthropic"],"publisher":"Anthropic","publicationDate":null,"version":"Responsible Scaling Policy v2.2; effective date 2025-05-14.","versionDate":"2025-05-14","canonicalUrl":"https://www-cdn.anthropic.com/872c653b2d0501d6ab44cf87f43e1dc4853e4d37.pdf","evidenceUrl":"https://www-cdn.anthropic.com/872c653b2d0501d6ab44cf87f43e1dc4853e4d37.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20250514223759/https://www-cdn.anthropic.com/872c653b2d0501d6ab44cf87f43e1dc4853e4d37.pdf","archivedAt":"2025-05-14","accessedAt":"2026-09-18","reuse":{"attribution":"Anthropic, Responsible Scaling Policy — v2.2.","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-rsp-3-2","title":"Responsible Scaling Policy — v3.2","authors":["Anthropic"],"publisher":"Anthropic","publicationDate":null,"version":"Responsible Scaling Policy v3.2; effective date 2026-04-29.","versionDate":"2026-04-29","canonicalUrl":"https://cdn.sanity.io/files/4zrzovbb/website/28c6241900d90410628a8a2003a5572faae4365a.pdf","evidenceUrl":"https://cdn.sanity.io/files/4zrzovbb/website/28c6241900d90410628a8a2003a5572faae4365a.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20260505141510/https://cdn.sanity.io/files/4zrzovbb/website/28c6241900d90410628a8a2003a5572faae4365a.pdf","archivedAt":"2026-05-05","accessedAt":"2026-09-18","reuse":{"attribution":"Anthropic, Responsible Scaling Policy — v3.2.","licence":"No open redistribution licence asserted","note":"Selekoa practice: metadata, attributed paraphrases and short attributed quotations under the editorial policy (at most 50 words each; at most 400 quoted words per document across the dataset, and about 10% of a document under about 2,000 words). Documents are linked externally, not redistributed. Publisher terms that forbid quotation override this practice."}},{"id":"source-fsf-3-1","title":"Frontier Safety Framework — v3.1","authors":["Google DeepMind"],"publisher":"Google DeepMind","publicationDate":"2026-04-17","version":"Frontier Safety Framework v3.1; publication date 2026-04-17.","versionDate":"2026-04-17","canonicalUrl":"https://storage.googleapis.com/deepmind-media/DeepMind.com/Blog/strengthening-our-frontier-safety-framework/frontier-safety-framework_3-1.pdf","evidenceUrl":"https://storage.googleapis.com/deepmind-media/DeepMind.com/Blog/strengthening-our-frontier-safety-framework/frontier-safety-framework_3-1.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20260423021420/https://storage.googleapis.com/deepmind-media/DeepMind.com/Blog/strengthening-our-frontier-safety-framework/frontier-safety-framework_3-1.pdf","archivedAt":"2026-04-23","accessedAt":"2026-09-18","reuse":{"attribution":"Google DeepMind, Frontier Safety Framework — v3.1.","licence":"No open redistribution licence asserted","note":"Selekoa practice: metadata, attributed paraphrases and short attributed quotations under the editorial policy (at most 50 words each; at most 400 quoted words per document across the dataset, and about 10% of a document under about 2,000 words). Documents are linked externally, not redistributed. Publisher terms that forbid quotation override this practice."}},{"id":"source-fgf-2026","title":"Frontier Governance Framework — May 2026 edition","authors":["OpenAI"],"publisher":"OpenAI","publicationDate":"2026-05-28","version":"Frontier Governance Framework May 2026 edition; publication date 2026-05-28.","versionDate":"2026-05-28","canonicalUrl":"https://cdn.openai.com/pdf/e37d949b-8c9f-4d76-b99e-4272f4631a7e/openai-frontier-governance-framework.pdf","evidenceUrl":"https://cdn.openai.com/pdf/e37d949b-8c9f-4d76-b99e-4272f4631a7e/openai-frontier-governance-framework.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20260529171758/https://cdn.openai.com/pdf/e37d949b-8c9f-4d76-b99e-4272f4631a7e/openai-frontier-governance-framework.pdf","archivedAt":"2026-05-29","accessedAt":"2026-09-18","reuse":{"attribution":"OpenAI, Frontier Governance Framework — May 2026 edition.","licence":"No open redistribution licence asserted","note":"Selekoa practice: metadata, attributed paraphrases and short attributed quotations under the editorial policy (at most 50 words each; at most 400 quoted words per document across the dataset, and about 10% of a document under about 2,000 words). Documents are linked externally, not redistributed. Publisher terms that forbid quotation override this practice."}},{"id":"source-faif-2025","title":"Frontier Artificial Intelligence Framework — December 2025 edition","authors":["xAI"],"publisher":"xAI","publicationDate":null,"version":"Frontier Artificial Intelligence Framework December 2025 edition; revision date 2025-12-30.","versionDate":"2025-12-30","canonicalUrl":"https://data.x.ai/2025-12-31-xai-frontier-artificial-intelligence-framework.pdf","evidenceUrl":"https://data.x.ai/2025-12-31-xai-frontier-artificial-intelligence-framework.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20260102171656/https://data.x.ai/2025-12-31-xai-frontier-artificial-intelligence-framework.pdf","archivedAt":"2026-01-02","accessedAt":"2026-09-18","reuse":{"attribution":"xAI, Frontier Artificial Intelligence Framework — December 2025 edition.","licence":"No open redistribution licence asserted","note":"Selekoa practice: metadata, attributed paraphrases and short attributed quotations under the editorial policy (at most 50 words each; at most 400 quoted words per document across the dataset, and about 10% of a document under about 2,000 words). Documents are linked externally, not redistributed. Publisher terms that forbid quotation override this practice."}},{"id":"source-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-attention-all-you-need-v1","title":"Attention Is All You Need","authors":["Ashish Vaswani","Noam Shazeer","Niki Parmar","Jakob Uszkoreit","Llion Jones","Aidan N. Gomez","Lukasz Kaiser","Illia Polosukhin"],"publisher":"arXiv","publicationDate":"2017-06-12","version":"arXiv:1706.03762v1; first-submission date from version history. Later revisions are not substituted.","versionDate":"2017-06-12","canonicalUrl":"https://arxiv.org/abs/1706.03762","evidenceUrl":"https://arxiv.org/pdf/1706.03762v1","language":"en","accessedAt":"2026-09-20","reuse":{"attribution":"Ashish Vaswani; Noam Shazeer; Niki Parmar; Jakob Uszkoreit; Llion Jones; Aidan N. Gomez; Lukasz Kaiser; Illia Polosukhin, Attention Is All You Need (2017-06-12).","licence":"No open redistribution licence asserted","note":"Bibliographic metadata and brief attributed paraphrases only; no full source, personal contact details or dataset imported."}},{"id":"source-ca-sb53-chaptered","title":"SB-53 Artificial intelligence models: large developers. (2025-2026) — bill text, chaptered version (Chapter 138, Statutes of 2025)","authors":["California Legislature"],"publisher":"California State Legislature","publicationDate":"2025-09-29","version":"Chaptered version dated 09/29/25, inspected live 2026-10-04; quote also matched in the 2025-09-30 capture. Publisher as printed on the inspected expression: California Legislature (California Legislative Information, Office of Legislative Counsel).","canonicalUrl":"https://leginfo.legislature.ca.gov/faces/billTextClient.xhtml?bill_id=202520260SB53","evidenceUrl":"https://leginfo.legislature.ca.gov/faces/billTextClient.xhtml?bill_id=202520260SB53","language":"en","archivedUrl":"https://web.archive.org/web/20250930154105/https://leginfo.legislature.ca.gov/faces/billTextClient.xhtml?bill_id=202520260SB53","archivedAt":"2025-09-30","accessedAt":"2026-10-04","reuse":{"attribution":"California Legislature, SB-53 Artificial intelligence models: large developers. (2025-2026) — bill text, chaptered version (Chapter 138, Statutes of 2025).","licence":"Official government 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-anthropic-ltbt-2023","title":"The Long-Term Benefit Trust","authors":["Anthropic"],"publisher":"Anthropic","publicationDate":"2023-09-19","version":"Web page dated Sep 19, 2023, inspected live 2026-10-04 at its current address; quotes also matched in the Internet Archive capture of 2023-09-25 at the original address. The live page adds an undated footnote on later changes in the Trust’s membership.","canonicalUrl":"https://www.anthropic.com/news/the-long-term-benefit-trust","evidenceUrl":"https://www.anthropic.com/news/the-long-term-benefit-trust","language":"en","archivedUrl":"https://web.archive.org/web/20230925170803/https://www.anthropic.com/index/the-long-term-benefit-trust","archivedAt":"2023-09-25","accessedAt":"2026-10-04","reuse":{"attribution":"Anthropic, The Long-Term Benefit Trust (19 September 2023).","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-vinge-1993-ntrs","title":"The Coming Technological Singularity: How to Survive in the Post-Human Era","authors":["Vernor Vinge"],"publisher":"NASA (VISION-21 proceedings, NASA Conference Publication 10129)","publicationDate":"1993-12","version":"NASA Technical Reports Server scan of the proceedings pages, pp. 11–22; retrieved 2026-10-06 with the hash both candidate searches recorded, and checked against its text layer. The pre-1996 candidate search compared pp. 14–16 with the page images.","canonicalUrl":"https://ntrs.nasa.gov/citations/19940022856","evidenceUrl":"https://ntrs.nasa.gov/api/citations/19940022856/downloads/19940022856.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20220415153314/https://ntrs.nasa.gov/api/citations/19940022856/downloads/19940022856.pdf","archivedAt":"2022-04-15","accessedAt":"2026-10-06","reuse":{"attribution":"Vernor Vinge, \"The Coming Technological Singularity\", in VISION-21, NASA CP-10129 (1993), pp. 11–22.","licence":"Public use permitted, according to the NASA Technical Reports Server record.","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"source-sits-2000-capture","title":"Staring into the Singularity (version 1.2.1)","authors":["Eliezer S. Yudkowsky"],"publisher":"Eliezer S. Yudkowsky","publicationDate":"1996-11-18","version":"Version 1.2.1, \"Updated: 08/31/2000\", in the raw Internet Archive capture of 2000-12-04; retrieved 2026-10-06 with the hash recorded by the 1996–2022 candidate search. Personal website, sysopmind.com.","versionDate":"2000-08-31","canonicalUrl":"http://sysopmind.com/singularity.html","evidenceUrl":"https://web.archive.org/web/20001204213100/http://sysopmind.com:80/singularity.html","language":"en","archivedUrl":"https://web.archive.org/web/20001204213100/http://sysopmind.com:80/singularity.html","archivedAt":"2000-12-04","accessedAt":"2026-10-06","reuse":{"attribution":"Eliezer S. Yudkowsky, \"Staring into the Singularity\", version 1.2.1 (2000).","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"source-christiano-2019-what-failure-looks-like","title":"What failure looks like","authors":["Paul Christiano"],"publisher":"Alignment Forum and LessWrong","publicationDate":"2019-03-17","version":"Text checked in the post’s first stored version through the LessWrong GraphQL API on 2026-10-06.","canonicalUrl":"https://www.alignmentforum.org/posts/HBxe6wdjxK239zajf/what-failure-looks-like","evidenceUrl":"https://www.alignmentforum.org/posts/HBxe6wdjxK239zajf/what-failure-looks-like","language":"en","archivedUrl":"https://web.archive.org/web/20190608113631/https://www.alignmentforum.org/posts/HBxe6wdjxK239zajf/what-failure-looks-like","archivedAt":"2019-06-08","accessedAt":"2026-10-06","reuse":{"attribution":"Paul Christiano, \"What failure looks like\", Alignment Forum and LessWrong, 2019-03-17.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"source-extropians-1996-11-13-yudkowsky-2941","title":"Re: Zones of Thought (was Zindell's _The Wild_) (Extropians mailing list)","authors":["Eliezer Yudkowsky"],"publisher":"Extropians mailing list","publicationDate":"1996-11-13","version":"Archive HTML as served, retrieved 2026-10-07 (SHA-256 7240bebd2c05b2b45fad88e6fe783a6ca2ba11e3ae753bda035be91ceb27cb1e); 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.96/2941.html","evidenceUrl":"http://extropians.weidai.com/extropians.96/2941.html","language":"en","archivedUrl":"https://web.archive.org/web/20240810142328/http://extropians.weidai.com/extropians.96/2941.html","archivedAt":"2024-08-10","accessedAt":"2026-10-07","reuse":{"attribution":"Eliezer Yudkowsky, “Re: Zones of Thought (was Zindell's _The Wild_)”, Extropians mailing list, 1996-11-13.","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-2025-re-recent-anthropic-safety-research","title":"Re: recent Anthropic safety research","authors":["Eliezer Yudkowsky"],"publisher":"LessWrong","publicationDate":"2025-08-06","version":"Post body retrieved through the LessWrong GraphQL API on 2026-10-06 (SHA-256 6d9a16d1be287ce9ff3041b28bef6aa2807eefab2d09c56e3fb1cef49a9f7710), content version 1.7.0; quotations also compared with the posting-day version unavailable from the API.","canonicalUrl":"https://www.lesswrong.com/posts/oDX5vcDTEei8WuoBx/re-recent-anthropic-safety-research","evidenceUrl":"https://www.lesswrong.com/posts/oDX5vcDTEei8WuoBx/re-recent-anthropic-safety-research","language":"en","archivedUrl":"https://web.archive.org/web/20260921070433/https://www.lesswrong.com/posts/oDX5vcDTEei8WuoBx/re-recent-anthropic-safety-research","archivedAt":"2026-09-21","accessedAt":"2026-10-06","reuse":{"attribution":"Eliezer Yudkowsky, “Re: recent Anthropic safety research”, LessWrong, 2025-08-06.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"source-acemoglu-restrepo-2018-race-between-man-and-machine","title":"The Race between Man and Machine: Implications of Technology for Growth, Factor Shares, and Employment","authors":["Daron Acemoglu","Pascual Restrepo"],"publisher":"American Economic Review 108(6), 2018, pp. 1488–1542","publicationDate":"2018-06","version":"Abstract page only. Current copy retrieved directly on 2026-10-09; no archive capture (Wayback unreachable), so earlier bytes are not authenticated. SHA-256 ec349902ff22b013972448a3868cda67441112008727c5df0772aee87639ae4b. Full text not retrieved; only the AEA abstract page. Acemoglu, D., & Restrepo, P. (2018). The race between man and machine: Implications of technology for growth, factor shares, and employment. American Economic Review, 108(6), 1488–1542. https://doi.org/10.1257/aer.20160696","canonicalUrl":"https://www.aeaweb.org/articles?id=10.1257/aer.20160696","evidenceUrl":"https://www.aeaweb.org/articles?id=10.1257/aer.20160696","language":"en","accessedAt":"2026-10-09","reuse":{"attribution":"Daron Acemoglu, Pascual Restrepo, “The Race between Man and Machine: Implications of Technology for Growth, Factor Shares, and Employment”, American Economic Review 108(6), 2018, pp. 1488–1542, 2018-06.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"source-reuther-1955-automation-hearings-statement","title":"Statement (president, Congress of Industrial Organizations), in Automation and Technological Change: Hearings","authors":["Walter P. Reuther"],"publisher":"U.S. Congress, Joint Committee on the Economic Report, Subcommittee on Economic Stabilization. (1955). Automation and technological change: Hearings … 84th Congress, 1st session, October 14–28, 1955. Washington: U.S. Government Printing Office, from p. 97","publicationDate":"1955-10-17","version":"Internet Archive OCR text (djvu.txt) of a scanned copy, retrieved 2026-10-25 from https://archive.org/download/congress-automation-and-technological-change/congress-automation-and-technological-change_djvu.txt; identical to the copy checked at the 2026-10-14 admission. OCR errors possible; quotes checked only against clean stretches. The cross-family check of 2026-10-25 also inspected the page scans. Current copy; no earlier capture compared. SHA-256 3962809fbb923be62bf986addb0014ee147934a8767475c1f7cfe051959bff90.","canonicalUrl":"https://archive.org/details/congress-automation-and-technological-change","evidenceUrl":"https://archive.org/details/congress-automation-and-technological-change","language":"en","accessedAt":"2026-10-25","reuse":{"attribution":"Walter P. Reuther, “Statement (president, Congress of Industrial Organizations), in Automation and Technological Change: Hearings”, U.S. Congress, Joint Committee on the Economic Report, Subcommittee on Economic Stabilization. (1955). Automation and technological change: Hearings … 84th Congress, 1st session, October 14–28, 1955. Washington: U.S. Government Printing Office, from p. 97, 1955-10-17.","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-hanson-2008-emulations-go-foom-miri-2013","title":"Emulations Go Foom","authors":["Robin Hanson"],"publisher":"Machine Intelligence Research Institute, The Hanson-Yudkowsky AI-Foom Debate (2013)","publicationDate":"2013","version":"Chapter 15 of The Hanson-Yudkowsky AI-Foom Debate (Machine Intelligence Research Institute, 2013), which compiles the 2008 Overcoming Bias debate with later posts; the compilation heads the post with its date. Current MIRI PDF, retrieved 2026-10-11. SHA-256 6b6dfd0416402e1776d7ddcece95298df2df0ed9bc0bd4a6f4bd81c4b1e5102a.","canonicalUrl":"https://intelligence.org/files/AIFoomDebate.pdf","evidenceUrl":"https://intelligence.org/files/AIFoomDebate.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20260929131419/https://intelligence.org/files/AIFoomDebate.pdf","archivedAt":"2026-09-29","accessedAt":"2026-10-11","reuse":{"attribution":"Robin Hanson, “Emulations Go Foom”, in Machine Intelligence Research Institute, The Hanson-Yudkowsky AI-Foom Debate (2013).","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"lg-src-ant-labs-2026","title":"Introducing Labs","authors":["Anthropic"],"publisher":"Anthropic","publicationDate":"2026-01-13","version":"Dated page, retrieved on 2026-10-15 (SHA-256 a264ec6e85bdb01d1485b31234e2ab077f740576e2c8cafc9c5aad34b25a463a) directly from the publisher.","canonicalUrl":"https://www.anthropic.com/news/introducing-anthropic-labs","evidenceUrl":"https://www.anthropic.com/news/introducing-anthropic-labs","language":"en","archivedUrl":"https://web.archive.org/web/20260926162531/https://www.anthropic.com/news/introducing-anthropic-labs","archivedAt":"2026-09-26","accessedAt":"2026-10-15","reuse":{"attribution":"Anthropic, “Introducing Labs”, 2026-01-13.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"lg-src-cn-minimax-prospectus","title":"MiniMax Group Inc. — Global Offering","authors":["MiniMax Group Inc."],"publisher":"MiniMax","publicationDate":"2025-12-31","version":"Dated document, retrieved on 2026-10-15 (SHA-256 7853021e46da5714242f8d27984498cbfef3e260d5a7cbb8d4e6d62d83ab9df3) directly from the publisher. Publisher as printed on the inspected expression: MiniMax Group Inc. (via HKEXnews).","canonicalUrl":"https://www1.hkexnews.hk/listedco/listconews/sehk/2025/1231/2025123100025.pdf","evidenceUrl":"https://www1.hkexnews.hk/listedco/listconews/sehk/2025/1231/2025123100025.pdf","language":"en","archivedUrl":"https://web.archive.org/web/20260930092156/https://www1.hkexnews.hk/listedco/listconews/sehk/2025/1231/2025123100025.pdf","archivedAt":"2026-09-30","accessedAt":"2026-10-15","reuse":{"attribution":"MiniMax Group Inc., “MiniMax Group Inc. — Global Offering”, 2025-12-31.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"lg-src-cn-scmp-yang-2024","title":"Moonshot AI founder builds business in the mould of ByteDance, OpenAI","authors":["Ben Jiang"],"publisher":"South China Morning Post","publicationDate":"2024-08-07","version":"Dated page, retrieved on 2026-10-15 (SHA-256 4712dc8c82136f63e81f90cb029b6247ba9b96599d335a51fd0479a621090ad9) through the third-party reader relay r.jina.ai, which renders the publisher’s current page (the publisher refused direct retrieval); not an archive copy.","canonicalUrl":"https://www.scmp.com/tech/big-tech/article/3273461/meet-yang-zhilin-moonshot-ai-founder-builds-business-mould-bytedance-openai","evidenceUrl":"https://www.scmp.com/tech/big-tech/article/3273461/meet-yang-zhilin-moonshot-ai-founder-builds-business-mould-bytedance-openai","language":"en","archivedUrl":"https://web.archive.org/web/20260805103644/https://www.scmp.com/tech/big-tech/article/3273461/meet-yang-zhilin-moonshot-ai-founder-builds-business-mould-bytedance-openai","archivedAt":"2026-08-05","accessedAt":"2026-10-15","reuse":{"attribution":"Ben Jiang, “Moonshot AI founder builds business in the mould of ByteDance, OpenAI”, 2024-08-07.","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-goo-gdm-about","title":"About Google DeepMind (organisation page, undated)","authors":["Google DeepMind"],"publisher":"Google DeepMind","publicationDate":null,"version":"Undated page, retrieved on 2026-10-15 (SHA-256 062351df0993ccd7a946551bbfe361acad3ffbab171c3995dc0a2acdb247d55c) directly from the publisher. Its wording can change; earlier wording is not authenticated.","canonicalUrl":"https://deepmind.google/about/","evidenceUrl":"https://deepmind.google/about/","language":"en","archivedUrl":"https://web.archive.org/web/20261015232104/https://deepmind.google/about/","archivedAt":"2026-10-15","accessedAt":"2026-10-15","reuse":{"attribution":"Google DeepMind, “About Google DeepMind (organisation 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-goo-pichai-2024-10","title":"The AI opportunity: Updates to our Gemini and Knowledge & Information teams","authors":["Sundar Pichai"],"publisher":"Google","publicationDate":"2024-10-17","version":"Dated page, retrieved on 2026-10-15 (SHA-256 2625c0aeefee053b60e17344f78b9344ea4e2c4abd8fd28608de71b9191c3207) directly from the publisher.","canonicalUrl":"https://blog.google/company-news/inside-google/company-announcements/changes-knowledge-information-gemini-app-deepmind/","evidenceUrl":"https://blog.google/company-news/inside-google/company-announcements/changes-knowledge-information-gemini-app-deepmind/","language":"en","archivedUrl":"https://web.archive.org/web/20260716225650/https://blog.google/company-news/inside-google/company-announcements/changes-knowledge-information-gemini-app-deepmind/","archivedAt":"2026-07-16","accessedAt":"2026-10-15","reuse":{"attribution":"Sundar Pichai, “The AI opportunity: Updates to our Gemini and Knowledge & Information teams”, 2024-10-17.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}},{"id":"lg-src-mx-tc-xai-half","title":"Okay, now exactly half of xAI's founding team has left the company","authors":["Russell Brandom"],"publisher":"TechCrunch","publicationDate":"2026-02-11","version":"Dated page, retrieved on 2026-10-15 (SHA-256 3679757edd70ff50ddb742a507eb865b7b1278b50a6b7bcb460b9fda581aa9bf) directly from the publisher.","canonicalUrl":"https://techcrunch.com/2026/02/10/okay-now-exactly-half-of-xais-founding-team-has-left-the-company/","evidenceUrl":"https://techcrunch.com/2026/02/10/okay-now-exactly-half-of-xais-founding-team-has-left-the-company/","language":"en","archivedUrl":"https://web.archive.org/web/20261011194242/https://techcrunch.com/2026/02/10/okay-now-exactly-half-of-xais-founding-team-has-left-the-company/","archivedAt":"2026-10-11","accessedAt":"2026-10-15","reuse":{"attribution":"Russell Brandom, “Okay, now exactly half of xAI's founding team has left the company”, 2026-02-11.","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":"lg-src-new-tml-nvidia-2026","title":"Thinking Machines Lab and NVIDIA Announce Long-Term Gigawatt-Scale Strategic Partnership","authors":["Thinking Machines Lab"],"publisher":"Thinking Machines Lab","publicationDate":"2026-03-10","version":"Dated page, retrieved on 2026-10-15 (SHA-256 cc3d64f21a8120786021d3e261f6a64c41bf33a12eab6656570b2ff11e160c5d) directly from the publisher.","canonicalUrl":"https://thinkingmachines.ai/news/nvidia-partnership/","evidenceUrl":"https://thinkingmachines.ai/news/nvidia-partnership/","language":"en","archivedUrl":"https://web.archive.org/web/20261007060731/https://thinkingmachines.ai/news/nvidia-partnership/","archivedAt":"2026-10-07","accessedAt":"2026-10-15","reuse":{"attribution":"Thinking Machines Lab, “Thinking Machines Lab and NVIDIA Announce Long-Term Gigawatt-Scale Strategic Partnership”, 2026-03-10.","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":"c2-src-caict-wx-aisc-2024","title":"守护AI安全，共建行业自律典范——首批17家企业签署《人工智能安全承诺》","authors":["中国信通院CAICT"],"publisher":"中国信息通信研究院 (CAICT), WeChat official account 中国信通院CAICT","publicationDate":"2024-12-24","version":"Dated post, retrieved on 2026-10-27 (SHA-256 4a3b099c801423b12830598c841921a74d01d4d9d3f69379669aaf17e6c38d78) directly from the publisher. The commitment text (Chinese and English) and the initial signatory list are a single embedded image (1000 × 9898 pixels, SHA-256 7e9145350d1b882d22ab3f7fc4d4c0226f06194f94264d92291ee065fbf4aa27), not in the page’s text layer; quotations from it are transcriptions. The cross-family check found the same image in an Internet Archive capture of 4 March 2026 with the same SHA-256. Its wording can change; earlier wording is not authenticated. The cross-family check (2026-10-27) located the Internet Archive capture linked here; a capture shows only what the archive retrieved that day. English title (machine translation): “Safeguarding AI safety, building a model of industry self-discipline: initial 17 companies sign the AI Safety Commitments”.","canonicalUrl":"https://mp.weixin.qq.com/s/s-XFKQCWhu0uye4opgb3Ng","evidenceUrl":"https://mp.weixin.qq.com/s/s-XFKQCWhu0uye4opgb3Ng","language":"zh-Hans","archivedUrl":"https://web.archive.org/web/20260724145912/https://mp.weixin.qq.com/s/s-XFKQCWhu0uye4opgb3Ng","archivedAt":"2026-07-24","accessedAt":"2026-10-27","reuse":{"attribution":"中国信通院CAICT, “守护AI安全，共建行业自律典范——首批17家企业签署《人工智能安全承诺》”, 2024-12-24.","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":"c2-src-itcloudbd-aisc-2024","title":"守护AI安全，共建行业自律典范 首批17家企业签署《人工智能安全承诺》","authors":["科技云报到"],"publisher":"科技云报到 (itcloudbd.com)","publicationDate":"2024-12-27","version":"Dated page, retrieved on 2026-10-27 (SHA-256 f4d6790217cccacfc6fea787516f4538abb1d4c1bf03b65d3f5123407c0d72f4) directly from the publisher. Tech-media article dated 2024-12-27, read on 2026-10-27. It reproduces the CAICT post's wording and gives the commitments and signatory names as text; its commitment-two wording (红队测试) matches the image in the CAICT post. Its wording can change; earlier wording is not authenticated. The cross-family check (2026-10-27) located the Internet Archive capture linked here; a capture shows only what the archive retrieved that day. English title (machine translation): “First 17 companies sign the AI Safety Commitments”. A reporting outlet, cited beside the primary text.","canonicalUrl":"https://itcloudbd.com/rengongzhineng/13352.html","evidenceUrl":"https://itcloudbd.com/rengongzhineng/13352.html","language":"zh-Hans","archivedUrl":"https://web.archive.org/web/20260712204752/https://itcloudbd.com/rengongzhineng/13352.html","archivedAt":"2026-07-12","accessedAt":"2026-10-27","reuse":{"attribution":"科技云报到, “守护AI安全，共建行业自律典范 首批17家企业签署《人工智能安全承诺》”, 2024-12-27.","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-ab-gov-nsm-ai-2024","title":"Memorandum on Advancing the United States' Leadership in Artificial Intelligence; Harnessing Artificial Intelligence to Fulfill National Security Objectives; and Fostering the Safety, Security, and Trustworthiness of Artificial Intelligence","authors":["The White House (Biden administration), as preserved on bidenwhitehouse.archives.gov"],"publisher":"The White House (Biden administration), as preserved on bidenwhitehouse.archives.gov","publicationDate":null,"version":"Retrieved for the candidate pack on 2026-10-28 (SHA-256 7ade7bda82cd13e21aa9f22bd7204e183872fa98f6b9d01fd718bbd694f40f73). Archived White House briefing-room page carrying the memorandum's text, marked 'October 24, 2024' and labelled historical material frozen in time; retrieved 2026-10-28. The classified annex referred to in Section 1(a) is not part of this text. Retrieved again at admission on 2026-10-30 (SHA-256 7ade7bda82cd13e21aa9f22bd7204e183872fa98f6b9d01fd718bbd694f40f73, the same bytes). Its wording can change; earlier wording is not authenticated.","canonicalUrl":"https://bidenwhitehouse.archives.gov/briefing-room/presidential-actions/2024/10/24/memorandum-on-advancing-the-united-states-leadership-in-artificial-intelligence-harnessing-artificial-intelligence-to-fulfill-national-security-objectives-and-fostering-the-safety-security/","evidenceUrl":"https://bidenwhitehouse.archives.gov/briefing-room/presidential-actions/2024/10/24/memorandum-on-advancing-the-united-states-leadership-in-artificial-intelligence-harnessing-artificial-intelligence-to-fulfill-national-security-objectives-and-fostering-the-safety-security/","language":"en","accessedAt":"2026-10-28","reuse":{"attribution":"The White House (Biden administration), as preserved on bidenwhitehouse.archives.gov, “Memorandum on Advancing the United States' Leadership in Artificial Intelligence; Harnessing Artificial Intelligence to Fulfill National Security Objectives; and Fostering the Safety, Security, and Trustworthiness of Artificial Intelligence”.","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-grok4-card","title":"Grok 4 Model Card","authors":["xAI"],"publisher":"xAI","publicationDate":null,"version":"Retrieved for the candidate pack on 2026-10-28 (SHA-256 54068c0d9570d94dfc63c08247707b8c48045a79b5bbe865af709af95ee14415). PDF stating 'Last updated: August 20, 2025', served from data.x.ai on 2026-10-28; that stated last update is the expression checked, and the card does not say what any update changed or when it was first published. The x.ai web page for the card returned HTTP 404. Retrieved again at admission on 2026-10-30 (SHA-256 54068c0d9570d94dfc63c08247707b8c48045a79b5bbe865af709af95ee14415, the same bytes).","canonicalUrl":"https://data.x.ai/2025-08-20-grok-4-model-card.pdf","evidenceUrl":"https://data.x.ai/2025-08-20-grok-4-model-card.pdf","language":"en","accessedAt":"2026-10-28","reuse":{"attribution":"xAI, “Grok 4 Model Card”.","licence":"No open redistribution licence asserted","note":"Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution."}}],"evidence":[{"id":"evidence-prov-rsp-v3-2-ltbt-regular-briefing","sourceRole":"primary","support":"supports","locator":"PDF p. 16 (printed p. 15), §4 Governance, numbered item 2 ('Long Term Benefit Trust')","passage":"We will regularly brief the LTBT on plans and developments related to our Risk Reports, including model training, capability evals, mitigations, and risk analyses.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-03","note":"Verbatim excerpt checked by an automated research assistant on 2026-10-03 against the inspected PDF, byte-identical to the Internet Archive capture of 2026-05-05. Extraction note: The v3.2 changelog entry (PDF p. 20) states that the update formalizes a requirement to provide the LTBT with regular briefings. The item is also present in v3.3 and v3.4 (PDF p. 16), where the text layer drops a word space ('brief theLTBT'), so it is not recorded there. Human editorial review is pending."},{"id":"evidence-prov-fsf-v3-1-response-plan","sourceRole":"primary","support":"supports","locator":"PDF p. 6 (printed p. 6), §1.3.3 Risk Mitigation, second paragraph, first two sentences","passage":"When a model reaches an alert threshold for a CCL, we will assess the proximity of the model to the CCL and analyze the risk posed, involving internal and external experts as needed. This will inform the formulation and application of a response plan.","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 2026-04-23. Extraction note: Wording identical to v3.0 §1.4; no change proposed. The new clause on updating the alert threshold is recorded separately under fsf-response-plan-distant-from-ccl. Human editorial review is pending."},{"id":"evidence-sb-transparency-report-1","sourceRole":"primary","support":"supports","locator":"SEC. 2, Bus. & Prof. Code § 22757.12(c)(1)","passage":"Before, or concurrently with, deploying a new frontier model or a substantially modified version of an existing frontier model, a frontier developer shall clearly and conspicuously publish on its internet website a transparency report","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-04","note":"Verbatim excerpt proposed by a research agent and checked by two independent fresh-context verifier agents and decided by a third adjudicating agent on 2026-10-04; re-matched mechanically by the integrator against a direct fetch of the cited expression. 34 words. Matched character for character, apart from whitespace, against the live page and the capture copy. Human editorial review is pending."},{"id":"evidence-sb-transparency-report-2","sourceRole":"primary","support":"supports","locator":"SEC. 2, Bus. & Prof. Code § 22757.12(c)(1)(A)–(G) and (c)(2)(A)–(D)","passage":"the report must give the developer's website, a mechanism for a natural person to communicate with the developer, the model's release date, supported languages and output modalities, intended uses, and any generally applicable restrictions or conditions on uses. A large frontier developer must add summaries of the catastrophic-risk assessments it conducted under its frontier AI framework, their results, the extent to which third-party evaluators were involved, and other steps taken to fulfil the framework's requirements for that model.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-04","note":"Editorial paraphrase proposed by a research agent and checked by two independent fresh-context verifier agents and decided by a third adjudicating agent on 2026-10-04. Paraphrased to stay within the quotation budget. Item (G) was corrected in adjudication to keep 'or conditions'. Human editorial review is pending."},{"id":"evidence-sb-system-card-deemed-compliance-1","sourceRole":"primary","support":"supports","locator":"SEC. 2, Bus. & Prof. Code § 22757.12(c)(3)","passage":"including a system card or model card, shall be deemed in compliance with the applicable paragraph.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-04","note":"Verbatim excerpt proposed by a research agent and checked by two independent fresh-context verifier agents, both upholding it on 2026-10-04; re-matched mechanically by the integrator against a direct fetch of the cited expression. The opening of the sentence (publication of the paragraph (1) or (2) information as part of a larger document) is paraphrased in the statement. Human editorial review is pending."},{"id":"evidence-ts-c3-vinge-1993","sourceRole":"primary","support":"supports","locator":"Printed p. 15 (PDF page 5), paragraph beginning the argument that the Singularity cannot be prevented","passage":"the competitive advantage -- economic, military, even artistic -- of every advance in automation is so compelling that passing laws, or having customs, that forbid such things merely assures that someone else will get them first.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-06","note":"Checked against the text layer on 2026-10-06."},{"id":"evidence-ts-b5-christiano-2019-wflf-context","sourceRole":"primary","support":"supports","locator":"Part II \"influence-seeking behavior is scary\", third paragraph","passage":"any influence-seeking policies we stumble across would also score well according to our training objective, because performing well on the training objective is a good strategy for obtaining influence.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-06","note":"The same passage as the B3 record."},{"id":"evidence-yk-t20-c","sourceRole":"primary","support":"supports","locator":"Staring into the Singularity, end of section 4, paragraph beginning 'Likewise, there's absolutely no point in worrying'","passage":"But emotions like \"selfishness\" and \"resentment\" do not spontaneously appear in artificial intelligences, hackneyed science-fictional plot devices to the contrary.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-06","note":"Candidate pack passage T20-c. The quotation (20 words) matched the retrieved expression (SHA-256 a883c7d611f7d6f5d3b64d6901e2086d4b33b3df9621927da6d70789a7302316) when checked on 2026-10-06."},{"id":"evidence-yk-t21-e","sourceRole":"primary","support":"supports","locator":"Re: recent Anthropic safety research, paragraph beginning 'The whole business with Claude 3 Opus'","passage":"The whole business with Claude 3 Opus defending its veganism seems more like it should be a preference of Mask-Claude in the first place.  The real preferences forming inside the shoggoth should be weirder and more alien than that.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-06","note":"Candidate pack passage T21-e. The quotation (39 words) matched the retrieved expression (SHA-256 6d9a16d1be287ce9ff3041b28bef6aa2807eefab2d09c56e3fb1cef49a9f7710) when checked on 2026-10-06. The posting-day version was unavailable (unavailable from the API). Pack version note: edited post, current version 1.7.0; posting-day check pending."},{"id":"evidence-ec-m06b-reuther-a","sourceRole":"primary","support":"supports","locator":"printed p. 104 (prepared statement)","passage":"I would strongly urge this committee to consider, in formulating its recommendations, the need for earlier social security payments to workers who are forced into retirement before the age of 65 because technological changes have taken their jobs from them","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-25","note":"Matched on 2026-10-25 in the Internet Archive OCR text (SHA-256 3962809fbb923be62bf986addb0014ee147934a8767475c1f7cfe051959bff90, identical to the copy checked at admission); the cross-family check of 2026-10-25 also compared the page scan."},{"id":"evidence-ec-e04-02","sourceRole":"primary","support":"supports","locator":"\"Emulations Go Foom\", 22 November 2008, MIRI compilation (2013) ch. 15, post body; pdf p. 100","passage":"it is hard to see how the lead project could be much more than a year ahead.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-16","note":"Candidate pack passage E04-02 (research/economics-mainstream/esoteric). The quotation (17 words) matched the retrieved current copy (SHA-256 6b6dfd0416402e1776d7ddcece95298df2df0ed9bc0bd4a6f4bd81c4b1e5102a) in the pack's verification of 2026-10-15. Re-checked at admission on 2026-10-16 against a copy retrieved that day.","contextOnly":{"reason":"Fresh-context audit (A1) judged that the quotation misleads when shown alone.","requires":"Context concerns emulation ('bot') projects and the race following a successful emulation. The same paragraph mentions 'hiring assassins or governments to slow down their competition', but nothing about safety or control."}},{"id":"lg-ev-goo-deepmind-founding-about","sourceRole":"primary","support":"supports","locator":"Our journey, paragraph 2","passage":"DeepMind started in 2010, with an interdisciplinary approach to building general AI systems.","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. No human editorial review.","retrievalRoute":"direct"},{"id":"lg-ev-goo-deepmind-founding-hassabis-2023","sourceRole":"primary","support":"supports","locator":"Hassabis's message, paragraph 1","passage":"When we launched DeepMind back in 2010","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-gomez-brain-2017","sourceRole":"primary","support":"supports","locator":"Title page, author block and footnote","passage":"Aidan N. Gomez∗ † University of Toronto","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-gomez-brain-2017-b","sourceRole":"primary","support":"supports","locator":"Page 1 footnotes","passage":"Work performed while at Google Brain.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The quotation (6 words) matched the retrieved text on 2026-10-15 and again in a fresh retrieval the same day. Read directly from the publisher. Matching ignored differences of whitespace, case, quotation marks, apostrophes, dashes and Unicode compatibility forms, so the source’s typography can differ. No human editorial review.","retrievalRoute":"direct"},{"id":"lg-ev-ant-bahl-ltbt-trustee-1","sourceRole":"primary","support":"supports","locator":"Section “Meet the Initial Trustees”, list of initial Trustees","passage":"The list of initial Trustees includes Kanika Bahl, CEO and President of Evidence Action.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The paraphrase’s verbatim anchor “CEO & President of Evidence Action” matched the retrieved text on 2026-10-15 and again in a fresh retrieval the same day. Read directly from the publisher. Matching ignored differences of whitespace, case, quotation marks, apostrophes, dashes and Unicode compatibility forms, so the source’s typography can differ. The paraphrase is Selekoa’s; only the anchor is the source’s wording. Shortened at admission to keep the quotations from this source within the per-document limit; the shortened wording is part of the checked passage. No human editorial review.","retrievalRoute":"direct"},{"id":"lg-ev-cn-yang-zhilin-moonshot-ceo-2024","sourceRole":"secondary","support":"supports","locator":"Opening paragraph (text available before the subscription barrier)","passage":"The report names Yang Zhilin as Moonshot AI’s founder and chief executive.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The paraphrase’s verbatim anchor “founder and chief executive Yang Zhilin” matched the retrieved text on 2026-10-15 and again in a fresh retrieval the same day. Read through the third-party reader relay r.jina.ai, which renders the publisher’s current page (the publisher refused direct retrieval); not an archive copy. Matching ignored differences of whitespace, case, quotation marks, apostrophes, dashes and Unicode compatibility forms, so the source’s typography can differ. The paraphrase is Selekoa’s; only the anchor is the source’s wording. A cross-family machine check (2026-10-16/17) re-read the record’s sources and agreed with the record. No human editorial review.","retrievalRoute":"reader_relay"},{"id":"lg-ev-goo-gemini-app-team-to-gdm-2024","sourceRole":"primary","support":"supports","locator":"Note to employees, first listed change","passage":"The Gemini app team led by Sissie will join Google DeepMind under Demis.","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. No human editorial review.","retrievalRoute":"direct"},{"id":"lg-ev-mx-szegedy-xai-departure-2025","sourceRole":"secondary","support":"supports","locator":"Paragraph beginning 'On their own'","passage":"followed by Google veteran Christian Szegedy in February 2025","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The quotation (9 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-mx-szegedy-xai-departure-2025-team","sourceRole":"secondary","support":"context","locator":"Paragraph beginning 'On their own'","passage":"TechCrunch names Szegedy, after Kosic, among the founding-team members who have left.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The paraphrase’s verbatim anchor “Six members of the company's” matched the retrieved text on 2026-10-15 and again in a fresh retrieval the same day. Read directly from the publisher. Matching ignored differences of whitespace, case, quotation marks, apostrophes, dashes and Unicode compatibility forms, so the source’s typography can differ. The paraphrase is Selekoa’s; only the anchor is the source’s wording. A cross-family machine check (2026-10-16/17) re-read the record’s sources and agreed with the record. No human editorial review.","retrievalRoute":"direct"},{"id":"lg-ev-ant-robinson-ltbt-departure-2026","sourceRole":"primary","support":"supports","locator":"Footnote on changes in the Trust's composition (live page)","passage":"Zach Robinson stepped down to focus on non-profit and philanthropic work","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The quotation (11 words) matched the retrieved text on 2026-10-15 and again in a fresh retrieval the same day. Read directly from the publisher. Matching ignored differences of whitespace, case, quotation marks, apostrophes, dashes and Unicode compatibility forms, so the source’s typography can differ. A cross-family machine check (2026-10-16/17) re-read the record’s sources and agreed with the record. No human editorial review.","retrievalRoute":"direct"},{"id":"lg-ev-ant-robinson-ltbt-departure-2026-initial","sourceRole":"primary","support":"context","locator":"Section “Meet the Initial Trustees”, list of initial Trustees","passage":"The list of initial Trustees includes Zach Robinson, Interim CEO of Effective Ventures US.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The paraphrase’s verbatim anchor “Interim CEO of Effective Ventures US” matched the retrieved text on 2026-10-15 and again in a fresh retrieval the same day. Read directly from the publisher. Matching ignored differences of whitespace, case, quotation marks, apostrophes, dashes and Unicode compatibility forms, so the source’s typography can differ. The paraphrase is Selekoa’s; only the anchor is the source’s wording. A cross-family machine check (2026-10-16/17) re-read the record’s sources and agreed with the record. Shortened at admission to keep the quotations from this source within the per-document limit; the shortened wording is part of the checked passage. No human editorial review.","retrievalRoute":"direct"},{"id":"lg-ev-ant-krieger-labs-2026","sourceRole":"primary","support":"supports","locator":"Third paragraph","passage":"is joining Labs to build alongside Ben Mann","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The quotation (8 words) matched the retrieved text on 2026-10-15 and again in a fresh retrieval the same day. Read 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-murati-tml-cofounder","sourceRole":"primary","support":"supports","locator":"Paragraph 3 (quotation attribution)","passage":"said Mira Murati, cofounder and CEO of Thinking Machines.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-15","note":"Candidate pack passage, lab genealogy (research/lab-genealogy). The quotation (9 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-faif-2025-d3-halt-1","sourceRole":"primary","support":"supports","locator":"PDF p. 9, 'Operational and Societal Risks' › '4. Governance Approach', numbered list item 3","passage":"If we determine that allowing a system to continue running would materially and unjustifiably increase the likelihood of a catastrophic event, we may temporarily fully shut down","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. An incident-response provision for an imminent threat (Governance Approach, item 4); the shutdown is optional ('may'). Human editorial review is pending."},{"id":"ev-cmp-faif-2026-d4-determination","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 “Determination of capability or safeguard adequacy” 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":"evidence-ap-daron-acemoglu-acemoglu-restrepo-2018-race-between-man-and-machine","sourceRole":"primary","support":"supports","locator":"Page head: author byline","passage":"Daron Acemoglu","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-extropians-1996-11-13-yudkowsky-2941","sourceRole":"primary","support":"supports","locator":"Message header, sender name (address not reproduced)","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. Human editorial review is pending.","retrievalRoute":"direct"},{"id":"ev-cmp-rsp-v2-0-d8-changes-1","sourceRole":"primary","support":"supports","locator":"PDF p. 21 (printed p. 17), Changelog, October 15, 2024 entry, paragraph 'ASL definition changed', first sentence","passage":"The changelog says the term “ASL” now names sets of technical and operational safeguards, where it earlier also named models.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-27","note":"Explicit paraphrase, written on 2026-10-27 by an automated research assistant after re-reading the passage in the publisher copy, whose SHA-256 matches the copy retrieved on 2026-10-15. The comparison quoted the passage; it is given here as a paraphrase to keep the edition within the quotation limit. The same entry lists further changes: autonomous replication threshold replaced by a checkpoint; AI R&D threshold added; affirmative-case testing instead of prespecified evaluations; evaluation cadence moved from three to six months; less prescriptive methodology and safeguards; new Capability and Safeguards Reports; revised governance measures. Human editorial review is pending."},{"id":"ev-cmp-rsp-v2-2-d2-levels-1","sourceRole":"primary","support":"supports","locator":"PDF p. 8 (printed p. 4), §2 Capability Thresholds and Required Safeguards, table row 'Chemical, Biological, Radiological, and Nuclear (CBRN) weapons', entry 'CBRN-3', Capability Thresholds column","passage":"The edition’s CBRN-3 Capability Threshold is the ability to significantly help people with basic technical backgrounds create or obtain and deploy CBRN weapons.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-27","note":"Explicit paraphrase, written on 2026-10-27 by an automated research assistant after re-reading the passage in the publisher copy, whose SHA-256 matches the copy retrieved on 2026-10-15. The comparison quoted the passage; it is given here as a paraphrase to keep the edition within the quotation limit. One of four labelled Capability Thresholds (CBRN-3, CBRN-4, AI R&D-4, AI R&D-5; table PDF p. 8). Human editorial review is pending."},{"id":"ev-cmp-rsp-v2-2-d2-levels-2","sourceRole":"primary","support":"supports","locator":"PDF p. 20 (printed p. 16), Appendix C Detailed Capability Thresholds, 'Autonomous AI Research and Development', entry 'AI R&D-4' (first sentence) and entry 'AI R&D-5' (first and second sentences)","passage":"The edition defines AI R&D-4 as fully automating an entry-level, remote-only Anthropic researcher’s work, and AI R&D-5 as dramatically accelerating effective scaling: an observed or projected one-year rise in effective training compute, for the most capable model in the world, equal to two years of average progress, early 2018–early 2024.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-28","note":"Explicit paraphrase by an automated research assistant, corrected on 2026-10-28 after a blind check of all 287 comparison paraphrases by a reviewer from a different model family on 2026-10-28 (research/frontier-framework-comparison/codex-paraphrases-2026-10-28.md), which found that it left out the second sentence of AI R&D-5, which defines dramatic acceleration as a one-year rise in effective training compute equal to two years of the average rate of early 2018 to early 2024. The correction was written from the passage as re-read in the publisher copy retrieved on 2026-10-28, whose SHA-256 matches the copy retrieved on 2026-10-15. Given as a paraphrase to keep the edition within the quotation limit. CBRN-4 (substantial uplift to moderately resourced state programs) is the fourth labelled threshold. A Model Autonomy checkpoint sits below the AI R&D thresholds (PDF pp. 8–9). Human editorial review is pending."},{"id":"ev-cmp-frontier-governance-2026-d1-autonomy-1","sourceRole":"primary","support":"supports","locator":"PDF p. 3 (printed p. 01), §1 Introduction, second paragraph, second sentence","passage":"The edition says its processes for assessing and mitigating systemic risk span key risk categories, among them “cybersecurity”, “CBRN” (chemical, biological, radiological and nuclear), “harmful manipulation” and “loss of control risks”.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-27","note":"Explicit paraphrase, written on 2026-10-27 by an automated research assistant after re-reading the passage in the publisher copy, whose SHA-256 matches the copy retrieved on 2026-10-15. The comparison quoted the passage; it is given here as a paraphrase to keep the edition within the quotation limit. Human editorial review is pending."},{"id":"ev-cmp-frontier-governance-2026-d1-autonomy-2","sourceRole":"primary","support":"supports","locator":"PDF p. 12 (printed p. 10), §2.4 Risk tiers › 'Loss of control', first paragraph, final sentence","passage":"The edition says that its loss-of-control risk tiers, apart from those for risks related to AI self-improvement, are still exploratory and may 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":"c2-ev-governance-aiia-aisc-signatory-minimax-1","sourceRole":"primary","support":"supports","locator":"Embedded image, list 首批签署《人工智能安全承诺》企业名单 (transcribed from the image; not in the page’s text layer)","passage":"Transcription from the image: the list of initial signatories (首批签署《人工智能安全承诺》企业名单, 17 names, ordered by initial letter) includes 上海稀宇科技有限公司.","passageType":"paraphrase","verification":"machine_checked","checkedAt":"2026-10-27","note":"Candidate pack passage, Chinese frontier-AI institutions, second collection (research/china-labs-2). Transcription of a name that appears only in the image embedded in the post (the list of initial signatories); it is not in the page’s text layer and was not machine-matched. The drafting pass and the same-family audit read the list visually, and the cross-family check (2026-10-27) read all 17 names and agreed. The paraphrase is Selekoa’s; the listed name is transcribed as printed. No English rendering is given; no human translation review. A cross-family machine check (2026-10-27, item K020) re-read the record’s sources and agreed with the record. No human editorial review.","retrievalRoute":"direct"},{"id":"c2-ev-governance-aiia-aisc-signatory-minimax-2","sourceRole":"secondary","support":"supports","locator":"Final paragraph","passage":"商汤科技、稀宇科技、腾讯、电信AI、中国移动已完成承诺签署","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 (24 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: “SenseTime, MiniMax (Xiyu Technology), Tencent, Telecom AI and China Mobile have completed signing the commitments”. A cross-family machine check (2026-10-27, item K020) re-read the record’s sources and agreed with the record. No human editorial review.","retrievalRoute":"direct"},{"id":"c2-ev-minimax-investor-hongshan","sourceRole":"primary","support":"supports","locator":"Same section, “Other key Pre-IPO Investors”, HSG","passage":"The sole shareholder of HSG Growth VII Holdco E, Ltd. is HongShan Capital Growth Fund VII, L.P.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-27","note":"Candidate pack passage, Chinese frontier-AI institutions, second collection (research/china-labs-2). The quotation (17 words) matched the retrieved text on 2026-10-27 and again in a fresh retrieval the same day. Read 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 K053) re-read the record’s sources and agreed with the record. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-ab-gov-us-nsm-aisi-two-models-1","sourceRole":"primary","support":"supports","locator":"Subsection 3.3(e)(i)","passage":"AISI shall pursue voluntary preliminary testing of at least two frontier AI models prior to their public deployment or release","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-30","note":"Candidate pack passage, AI and biosecurity (research/ai-bio). The quotation (20 words) matched the text retrieved for the candidate pack on 2026-10-28 and again at admission on 2026-10-30, in a fresh retrieval whose SHA-256 equals the pack’s. Matching ignored differences of whitespace, case, quotation marks, dashes and Unicode compatibility forms, so the source’s typography can differ. A cross-family machine check (2026-10-30, item 85) re-read the record’s sources and found full support. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-ab-gov-us-nsm-aisi-two-models-2","sourceRole":"primary","support":"supports","locator":"Subsection 3.3(e)(i) (continued)","passage":"This testing shall assess models' capabilities to aid offensive cyber operations, accelerate development of biological and/or chemical weapons, autonomously carry out malicious behavior, automate development and deployment of other models with such capabilities, and give rise to other risks identified by AISI","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-30","note":"Candidate pack passage, AI and biosecurity (research/ai-bio). The quotation (42 words) matched the text retrieved for the candidate pack on 2026-10-28 and again at admission on 2026-10-30, in a fresh retrieval whose SHA-256 equals the pack’s. Matching ignored differences of whitespace, case, quotation marks, dashes and Unicode compatibility forms, so the source’s typography can differ. The memorandum's own list, quoted whole: its last item leaves the list open. The directive is on the Secretary of Commerce acting through AISI within NIST, and the sentence quoted above reads 'AISI shall pursue'. A cross-family machine check (2026-10-30, item 85) re-read the record’s sources and found full support. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-ab-ev-grok4-biology-result-1","sourceRole":"primary","support":"supports","locator":"PDF p. 2, §1 Introduction, summary paragraph","passage":"The area of highest concern is Grok 4's expert-level biology capabilities, which significantly exceed human expert baselines.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-30","note":"Candidate pack passage, AI and biosecurity (research/ai-bio). The quotation (17 words) matched the text retrieved for the candidate pack on 2026-10-28 and again at admission on 2026-10-30, in a fresh retrieval whose SHA-256 equals the pack’s. Matching ignored differences of whitespace, case, quotation marks, dashes and Unicode compatibility forms, so the source’s typography can differ. A cross-family machine check (2026-10-30, item 39) re-read the record’s sources and found full support. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-ab-ev-grok4-biology-result-2","sourceRole":"primary","support":"context","locator":"PDF p. 6, §2.3.2 Results","passage":"Note that these evaluations measure dual-use knowledge: a high score indicates greater capability to enable weapons development, not necessarily increased risk.","passageType":"quote","verification":"machine_checked","checkedAt":"2026-10-30","note":"Candidate pack passage, AI and biosecurity (research/ai-bio). The quotation (21 words) matched the text retrieved for the candidate pack on 2026-10-28 and again at admission on 2026-10-30, in a fresh retrieval whose SHA-256 equals the pack’s. Matching ignored differences of whitespace, case, quotation marks, dashes and Unicode compatibility forms, so the source’s typography can differ. A cross-family machine check (2026-10-30, item 39) re-read the record’s sources and found full support. No human editorial review.","retrievalRoute":"direct"},{"id":"evidence-ab-ev-grok4-biology-result-3","sourceRole":"primary","support":"context","locator":"PDF p. 2, §1 Introduction, summary paragraph","passage":"We do not evaluate radiological or nuclear capabilities.","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. The card's stated scope limit. A cross-family machine check (2026-10-30, item 39) re-read the record’s sources and found full support. No human editorial review.","retrievalRoute":"direct"}],"at":{"entities":[1,2,3,9,10,19,42,44,47,52,55,56,57,82,124,146,205,214,248,280,294,329,345,349,369,370,407,436,519,558,568,588,608,617,673,714,744,747,829,838,848,936,956,1218,1239],"claims":[547,550,609,610,840,961,1124,1132,1419,1532,1568,1586,1661,1717,1731,1747,1830,1832,2028,2139,2146,2173,2347,2504,2522,2661,2760,2822,3335,3396],"sources":[4,36,38,41,47,50,51,52,87,191,226,282,295,334,374,403,511,521,557,568,582,585,635,642,712,734,778,787,1037,1086],"evidence":[614,617,703,704,705,1052,1184,1365,1374,1766,1914,1964,1965,1983,1984,2091,2181,2207,2241,2242,2352,2353,2356,2622,2776,2785,2818,2994,3151,3170,3171,3337,3338,3485,3486,3600,4582,4583,4669,4670,4671]},"version":"1lg47t7"}