AI Matters
Why it matters
Anthropic grants external auditors deep access to evaluate model risks
TechCrunch Anthropic is letting METR researchers look under the hood with employee-level access
Anthropic gave research organization METR employee-level access to assess AI safety risks without retaining the power to block published findings.
Why it matters
Independent oversight helps ensure safety issues and risks in advanced models are publicly disclosed without corporate censorship.
Latest brief · Oct 19, 2026