Center for AI Safety; retrieved 2026-09-15 tier 3 self (cais)confirmed
Research Projects | CAIS
Research Projects page; benchmarks (AgentHarm, WMDP, MASK) evaluate leading LLMs; no privileged access claimed. Fetched 2026-09-15.
Signals citing this source
- Center for AI Safety, A (against): Public-model access.“We evaluate a range of leading LLMs”
Ledger rows citing this URL
- none