SaferAI Ratings
by SaferAIPublished 2025-10-01Source ↗Overall:Very Weak
- Risk Analysis and Evaluation
- Very Weak
- Risk Governance
- Weak
- Risk Identification
- Very Weak
- Risk Treatment
- Weak
Grades from external scorecards. We mirror published grades only; per-source methodology lives at the link in each panel header. See the scorecards directory for cross-org comparison.
AI Lab Watch
by Zach Stein-PerlmanPublished 2025-09-01Source ↗Overall:Very Weak
- Risk info sharing
- Very Weak
- Misuse prevention
- Very Weak
- Planning
- Very Weak
- Risk assessment
- Very Weak
- Boosting safety research
- Very Weak
- Scheming risk prevention
- Very Weak
- Prep for extreme security
- Very Weak
Seoul Commitment Tracker
by The Midas ProjectPublished 2025-02-10Source ↗Overall:Fulfilled
- Halting Procedures
- Fulfilled
- Risk Evaluation
- Fulfilled
- Risk Mitigations
- Partial
- Risk Thresholds
- Partial
- Safety Investment
- Partial