Evaluator Bench
arXiv, 2026-01-21; retrieved 2026-09-14 tier 4 press (primary)confirmed

Expanding External Access to Frontier AI Models for Dangerous Capability Evaluations (arXiv 2601.11916)

https://arxiv.org/abs/2601.11916

Taxonomy of access levels for external evaluators. Cited from a search snippet; not re-fetched in full. [verify 2026-09-15] fetched; page reachable, citation matches

Signals citing this source

  • EU AI Office, S (for): Code sets the access floor: helpful-only versions where security allows, at least 20 business days, external evaluators for each new frontier model.

Ledger rows citing this URL

  • none