What it is
NIST AI 600-1, Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile, was released on 26 July 2024. It applies the AI RMF to generative AI and gives suggested actions coded to the RMF functions (GV, MP, MS, MG). Like the RMF, it is voluntary.
The 12 generative AI risks
- CBRN information or capabilities
- Confabulation
- Dangerous, violent, or hateful content
- Data privacy
- Environmental impacts
- Harmful bias or homogenization
- Human-AI configuration
- Information integrity
- Information security
- Intellectual property
- Obscene, degrading, and/or abusive content
- Value chain and component integration
For RAG systems, the most relevant are confabulation, information security (including prompt injection), data privacy, intellectual property, and value chain and component integration (third-party models and data).
Status
We found no revision of AI 600-1. NIST is revising the parent AI RMF under the July 2025 AI Action Plan, so expect the profile to be reviewed in line with it.
Reuse terms
As a NIST publication, it is a US Government work without copyright protection (17 U.S.C. 105) and is freely available at doi.org/10.6028/NIST.AI.600-1. Credit NIST, and check any third-party material cited inside.
Using the profile in AI and RAG
- Tag each suggested action with its risk and RMF function.
- Map your RAG controls (source grounding, citation, access control) to confabulation, privacy, and security risks.
- Track third-party components under value chain risk.
- Keep the profile linked to AI RMF 1.0 and note the parent revision.
- Label it voluntary in answers.
Common mistakes
- Treating the profile as a checklist that must be fully implemented.
- Ignoring IP and value chain risks in RAG.
- Citing action counts without checking the source.
- Assuming it replaces the AI RMF.
Checklist
- Actions tagged by risk and function
- RAG controls mapped to relevant risks
- Third-party components inventoried
- Link to AI RMF 1.0 and revision status
- Voluntary status labeled
How AskDona helps
AskDona grounds every answer in cited source passages, which directly addresses confabulation, and can check your GenAI policies against the profile’s risks with evidence for each.
Run a sample generative AI risk check
AskDona shows the evidence; your AI risk team decides.
Editor’s note. This article is general information, not legal advice. The “conditions to check” rating comes from our source list and is a first-pass judgment, not a legal opinion.
FAQ
What is NIST AI 600-1? The Generative AI Profile of the NIST AI RMF, released 26 July 2024.
How many risks does it list? Twelve.
Is it mandatory? No, it is voluntary.
Which risks matter most for RAG? Confabulation, information security, data privacy, intellectual property, and value chain integration.
References
All pages opened and checked by October 26, 2026.
| Source | What it supports |
|---|---|
| NIST AI 600-1 (PDF) | Date and 12 risks |
| NIST: AI Risk Management Framework | Release date, RMF revision |
| NIST: Copyrights and disclaimers | Reuse terms |
| 17 U.S.C. 105 (LII) | No copyright in US Government works |