Claim and evidence Close ×
Activating AI Safety Level 3 protections states evaluation result ↓ Anthropic states that it has not determined whether Claude Opus 4 passed the ASL-3 capabilities threshold, that it could not clearly rule out ASL-3 risks as it had for previous models, and that it deployed the model with ASL-3 measures as a precautionary and provisional action. Anthropic's ASL-3 activation post states that it had not yet determined whether Claude Opus 4 definitively passed the capabilities threshold requiring ASL-3 protections, that continued improvements in CBRN-related knowledge and capabilities meant clearly ruling out ASL-3 risks was not possible as it had been for every previous model, and that it deployed the model with ASL-3 measures as a precautionary and provisional action. Copy claim link
Copy claim reference
Machine-checked, not reviewed by an editor Stated in source Supported
Supporting evidence: primary · 1 source
Subject version As published on anthropic.com on 22 May 2025 · Published 2025-05-22 · Effective date not established Date 2025-05-22 · publication Evaluated system Claude Opus 4 (with the post stating that Claude Sonnet 4 was ruled out of needing ASL-3) What the report measured Whether Claude Opus 4 reaches the ASL-3 capability threshold, which concerns CBRN weapons misuse By what kind of method The developer's own dangerous-capability evaluations under its Responsible Scaling Policy The result, in the report’s words Not determined; ASL-3 risks could not be clearly ruled out, so ASL-3 Deployment and Security Standards were applied as a precautionary and provisional action, and the ASL-4 standard was ruled out The caveats the report gives The post states that dangerous capability evaluations are inherently challenging and that as models approach the thresholds it takes longer to determine their status. Scope and limits A developer's precautionary decision under its own policy, stated as provisional; the post says the protections may be removed or adjusted if it later concludes the threshold was not surpassed. It is not an independent finding about the model's capability and says nothing about whether the measures work.
The evidence 02 primary source Supports the claim
Activating AI Safety Level 3 protections ↗ Anthropic · Publication date unknown
Copy source reference
Cite the original source Cite the original document, not Selekoa. Check these details against it.
Copy APA
Download BibTeX Download RIS Download CSL-JSON
Source excerpt “we have not yet determined whether Claude Opus 4 has definitively passed the Capabilities Threshold that requires ASL-3 protections” Location in source First section, second paragraph
How it was read Retrieved directly from the publisher
Machine-checked · 2026-10-30
Source details and check notes Authors Anthropic Version Retrieved for the candidate pack on 2026-10-28 (SHA-256 411f66885bacd8769dd6718e78f40eb0df80a4983dbde9d0d49f0323746e8656). Page labelled 'Published May 22, 2025', with six footnotes, as served on 2026-10-28. The accompanying full report (a separate PDF) was not retrieved. Retrieved again at admission on 2026-10-30 (SHA-256 35fa989dfc4f36842ef4f1dda2cfbdda9f9fc9f63524b310fd869ed2f32ab1e8). Its wording can change; earlier wording is not authenticated. Canonical address Open the canonical source ↗ Accessed 2026-10-28 Language English (en) Attribution Anthropic, “Activating AI Safety Level 3 protections”. Reuse No open redistribution licence asserted. Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution. Check notes Candidate pack passage, AI and biosecurity (research/ai-bio). The quotation (19 words) matched the text retrieved for the candidate pack on 2026-10-28 and again at admission on 2026-10-30, in a fresh retrieval (SHA-256 35fa989dfc4f36842ef4f1dda2cfbdda9f9fc9f63524b310fd869ed2f32ab1e8; the bytes differ from the pack’s retrieval). Matching ignored differences of whitespace, case, quotation marks, dashes and Unicode compatibility forms, so the source’s typography can differ. A cross-family machine check (2026-10-30, item 29) re-read the record’s sources and found full support. No human editorial review. primary source Supports the claim
Activating AI Safety Level 3 protections ↗ Anthropic · Publication date unknown
Copy source reference
Cite the original source Cite the original document, not Selekoa. Check these details against it.
Copy APA
Download BibTeX Download RIS Download CSL-JSON
Paraphrase The post states that, because of continued improvements in CBRN-related knowledge and capabilities, clearly ruling out ASL-3 risks was not possible for Claude Opus 4 in the way it had been for every previous model, and that more detailed study is required.
Location in source First section, second paragraph
How it was read Retrieved directly from the publisher
Machine-checked · 2026-10-30
Source details and check notes Authors Anthropic Version Retrieved for the candidate pack on 2026-10-28 (SHA-256 411f66885bacd8769dd6718e78f40eb0df80a4983dbde9d0d49f0323746e8656). Page labelled 'Published May 22, 2025', with six footnotes, as served on 2026-10-28. The accompanying full report (a separate PDF) was not retrieved. Retrieved again at admission on 2026-10-30 (SHA-256 35fa989dfc4f36842ef4f1dda2cfbdda9f9fc9f63524b310fd869ed2f32ab1e8). Its wording can change; earlier wording is not authenticated. Canonical address Open the canonical source ↗ Accessed 2026-10-28 Language English (en) Attribution Anthropic, “Activating AI Safety Level 3 protections”. Reuse No open redistribution licence asserted. Bibliographic metadata, editorial paraphrases and short attributed quotations (at most 50 words each). Recheck terms before wider redistribution. Check notes Candidate pack passage, AI and biosecurity (research/ai-bio). The paraphrase is Selekoa’s, written from the text retrieved for the candidate pack on 2026-10-28; it is not a quotation and was not machine-matched. Paraphrased to keep within this short document's quotation budget. A cross-family machine check (2026-10-30, item 29) re-read the record’s sources and found full support. No human editorial review. Status and meaning Editorial status. Not reviewed by an editor . An exploratory record that no editor has accepted. This status does not say whether a person has read any linked passage. Each evidence record above shows how it was checked.
Relation. A report states the result of an evaluation it ran or commissioned, in its own words, with what was measured, by what kind of method, and the caveats the report itself gives (schema 2.22). The record is the report’s statement, not a verified finding: it says nothing about another system, a later version or real-world risk, and a null or limiting result is not evidence that no effect exists.
Date. The determination as stated on 22 May 2025.
Record ID ab-ev-anthropic-asl3-opus4-determination