Choose an institution to see its records and dossier, or choose a topic.
Through year 2027 · all records, including scheduled dates 2026 2025 2024 2023 2022 2021 2020 2019 2018 2017 2016 2015 2014 2013 2012 2011 2010 2008 2005 1965 Show Organisations and infrastructure All types Organisation Person Document Concept Infrastructure Public body Reset filters
94 people, organisations and other records · 122 relationships · 34 claims about a valueEverything recorded up to the chosen year, whether or not still in effect · Undated records included
Claim and evidence Close ×
v3.1 defines alert thresholds for CCLs that may draw on evaluation results, expert assessments and other sources of information and flag when a CCL may be reached before a critical capability assessment is conducted again; v3.0 defined alert thresholds for its early warning evaluations that flag when a CCL may be reached before a risk assessment is conducted again, with a separate note allowing the ML R&D alert threshold to be defined from information on Google DeepMind’s own ML R&D progress rather than evaluation scores. Copy claim link
Copy claim reference
Machine-checked, not reviewed by an editor Inferred Supported
Supporting evidence: primary · 2 sources
Reasoning and premises Premises: FSF · v3.1 states the provision as quoted in its provision record (PDF p. 5, §1.3.2 Inherent Risk Assessment, third paragraph, second sentence); FSF · v3.0 states it as quoted in its provision record (PDF p. 5, §1.3, “Analysis” bullet, second sentence; Note on Machine Learning R&D CCLs, PDF pp. 5–6, last sentence). The difference described is read directly from the two quoted passages.
Basis of the change Inferred by comparing the two editions. No publisher statement of this change is recorded. Provision Threshold or trigger · provision ID fsf-alert-threshold Subject version v3.1 · Published 2026-04-17 · Effective date not established Object version v3.0 · Published 2025-09-22 · Effective date not established Date 2026-04-17 · publication Scope and limits A textual difference between the two editions’ wording of this keyed provision. It is not an assessment of whether the change strengthened or weakened the framework, and not evidence of implementation.
The evidence 02 primary source Supports the claim
Frontier Safety Framework — v3.1 ↗ Google DeepMind · 2026-04-17
Archived copy, captured 2026-04-23: Frontier Safety Framework — v3.1 ↗
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 FSF v3.1 states that, for CCLs, Google DeepMind defines alert thresholds, which may draw on evaluation results, expert assessments and other sources of information, designed to flag when a CCL may be reached before a critical capability assessment is conducted again.
Location in source PDF p. 5, §1.3.2 Inherent Risk Assessment, third paragraph, second sentence
Machine-checked · 2026-10-03
Source details and check notes Authors Google DeepMind Version Frontier Safety Framework v3.1; publication date 2026-04-17. Version date 2026-04-17 Canonical address Open the canonical source ↗ Accessed 2026-09-18 Language English (en) Archive copy Open archived copy, captured 2026-04-23: Frontier Safety Framework — v3.1 ↗ . Captured 6 days after the recorded version date. An archive copy shows only what the archive retrieved on 2026-04-23. It does not prove what the page said earlier, and it can differ from the copy checked on 2026-09-18.Attribution Google DeepMind, Frontier Safety Framework — v3.1. Reuse No open redistribution licence asserted. 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. Check notes Paraphrase checked by an automated research assistant on 2026-10-03 against the cited edition; the edition’s own wording is quoted in its provision record. Human editorial review is pending. primary source Supports the claim
Frontier Safety Framework — v3.0 ↗ Google DeepMind · 2025-09-22
Archived copy, captured 2025-09-22: Frontier Safety Framework — v3.0 ↗
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 v3.0 designs its alert thresholds for early warning evaluations to flag when a CCL may be reached before a risk assessment is conducted again, and its Note on Machine Learning R&D CCLs lets the ML R&D alert threshold be defined from Google DeepMind’s own ML R&D progress information rather than evaluation scores.
Location in source PDF p. 5, §1.3, “Analysis” bullet, second sentence; Note on Machine Learning R&D CCLs, PDF pp. 5–6, last sentence
Machine-checked · 2026-10-03
Source details and check notes Authors Google DeepMind Version Frontier Safety Framework v3.0; publication date 2025-09-22. Version date 2025-09-22 Canonical address Open the canonical source ↗ Accessed 2026-09-18 Language English (en) Archive copy Open archived copy, captured 2025-09-22: Frontier Safety Framework — v3.0 ↗ . Captured on the recorded version date. An archive copy shows only what the archive retrieved on 2025-09-22. It does not prove what the page said earlier, and it can differ from the copy checked on 2026-09-18.Attribution Google DeepMind, Frontier Safety Framework — v3.0. Reuse No open redistribution licence asserted. 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. Check notes Paraphrase checked by an automated research assistant on 2026-10-03 against the cited edition; the edition’s own wording is quoted in its provision record. Human editorial review is pending. 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. One scoped difference between document versions. The claim identifies publisher-stated change, official changelog/redline, or an explicitly inferred comparison. It does not establish implementation, effectiveness or a safety ranking.
Date. Dated by the later edition’s publication date.
Record ID provchange-fsf-v3-1-alert-threshold