Every AI Policy Ever · open in the interactive register
Draft / proposedProposed · Guidelines / framework
Managing Misuse Risk for Dual-Use Foundation Models
Record US-07 · NIST AI 800-1
- Issued by
- U.S. AI Safety Institute, National Institute of Standards and Technology (NIST)
- Jurisdiction
- United States
- Level · region
- National · North America
- Published / adopted
- 2024-07
- Last updated
- 2025-01-15
- Language
- English
- Official identifier
- NIST AI 800-1
Summary
Draft voluntary guidelines of the U.S. AI Safety Institute at NIST, chiefly for initial developers of dual-use foundation models, on managing the risk that models are deliberately misused to harm public safety or national security, for example to develop chemical, biological, radiological or nuclear weapons, enable offensive cyber-attacks or generate child sexual abuse material. The second draft sets seven objectives: identify potential misuse risk, plan to manage it, protect the model from unauthorized access, measure it, mitigate it before deployment, monitor and respond to misuse, and disclose risk management practices.
Safety topics
general-purpose / frontier AIevaluation & testingcatastrophic risksafety institutetransparency & labelling
Status notes
Initial public draft released in July 2024 for public comment. The second public draft (NIST AI 800-1 2pd) was released on 15 January 2025, with comments open until 15 March 2025. It added an appendix on measuring misuse risk, appendices on chemical and biological and on cyber misuse risk, a 'marginal risk' framework, guidance for developers of open models and content for other actors across the AI supply chain. It is consistent with the National AI Initiative Act, Executive Order 14110 and the National Security Memorandum on AI of 24 October 2024, and builds on the AI Risk Management Framework and its Generative AI Profile.
Local copy
Archived. Last attempt 2026-10-17 22:11:57 UTC
Official source (third-party copy, not published) archive/US-07/2026-10-11/official.pdf
Retrieved 2026-10-11 14:03:45 UTC
SHA-256 30ced091493af421d454ca138a3496a39b8c6079f74223e868325318ffa767e4
Verification
AThe instrument's own official text, or an official register, was stored and confirms the title, date and status. checked 2026-10-18 · official-url-fetched
Recheck of 17 Oct 2026: re-read the initial public draft PDF on nvlpubs.nist.gov (stored 11 Oct 2026, unchanged on 17 Oct 2026), and stored in this run NIST's news release of 15 Jan 2025 and the second public draft (NIST AI 800-1 2pd, January 2025). Together they confirm the title, NIST AI 800-1, the U.S. AI Safety Institute as issuer, July 2024 for the first draft, 15 Jan 2025 for the second (date_last_updated), the draft status and the comment period to 15 Mar 2025. The summary now follows the second draft, the current text. That no final version exists rests on a web search on 17 Oct 2026 that found none and on version 1's reading of NIST's guidelines page; no stored file shows it. official_identifier is now the publication number alone, since each draft has its own DOI.
Second check: Independent check by Claude Code (claude-sonnet-5-5), 2026-10-18: 7 field(s) supported.
History
- 2026-10-18notes: Initial public draft July 2024; second public draft released for comment on 15 J → Initial public draft released in July 2024 for public comment. The second public — rewritten: both drafts, the comment period and what the second draft added (from NIST's news release), and the instruments it is consistent with (from the second draft)2026-10-17-recheck-north-america-1 · Claude Code (claude-opus-5-5)
- 2026-10-18related: [{"slug": "us-aisi-establishment", "id": "US-05"}, {"slug": "us-caisi", "id": "U → [{"id": "US-05", "slug": "us-aisi-establishment"}, {"id": "US-15", "slug": "us-c — US-02 and US-08 added (the second draft builds on the AI RMF and its Generative AI Profile); US-09 added (the National Security Memorandum on AI of 24 Oct 2024, which the second draft is consistent with)2026-10-17-recheck-north-america-1 · Claude Code (claude-opus-5-5)
- 2026-10-18summary: Draft voluntary guidance for foundation-model developers on managing deliberate → Draft voluntary guidelines of the U.S. AI Safety Institute at NIST, chiefly for — rewritten from the second public draft, the current text: 'voluntary guidelines', its harm to public safety or national security, and its seven objectives in its own words2026-10-17-recheck-north-america-1 · Claude Code (claude-opus-5-5)
- 2026-10-18official_identifier: NIST AI 800-1 ipd (DOI 10.6028/NIST.AI.800-1.ipd); second public draft NIST AI 8 → NIST AI 800-1 — 'second public draft NIST AI 800-1 2pd' is a description, which the field excludes; the publication number alone remains, since each draft has its own DOI (.ipd, .2pd)2026-10-17-recheck-north-america-1 · Claude Code (claude-opus-5-5)
Citation
U.S. AI Safety Institute, National Institute of Standards and Technology (NIST). Managing Misuse Risk for Dual-Use Foundation Models [NIST AI 800-1]. 2024-07. https://nvlpubs.nist.gov/nistpubs/ai/NIST.AI.800-1.ipd.pdf. Status: Draft / proposed; verified 2026-10-18 (level A). Local copy archive/US-07/2026-10-11/official.pdf retrieved 2026-10-11 14:03:45 UTC, SHA-256 30ced091493af421d454ca138a3496a39b8c6079f74223e868325318ffa767e4. Every AI Policy Ever, version 1.18 (snapshot 2026-10-18), record US-07.
Improve this record
Know of an amendment, a status change or an error? Report news or an error on GitHub; the form opens with this record filled in.