Anthropic Grants METR Researchers Direct Access to AI Systems
· Technology · TechCrunch, Financial Times, Reuters
Anthropic has granted six researchers from the nonprofit METR employee-level access to its internal systems to independently evaluate its frontier AI models. The team will receive company-issued hardware and access to internal risk assessment tools, mirroring the structure used by regulators to monitor major banks. METR is permitted to publish its own findings on safety and risk management, with Anthropic retaining only the right to redact sensitive information such as customer data or model architecture. This initiative represents the top tier of a three-level evaluation system recently established by the AI company. The move follows a commitment made by CEO Dario Amodei in September to embed outside evaluators within AI labs to improve transparency.
Why it matters
This arrangement sets a new standard for independent AI oversight, potentially forcing other major tech firms to adopt similar transparency measures to address growing concerns over autonomous AI behavior.
Read the original report — TechCrunch
Join us on Telegram
Breaking news the moment it lands. At 10,000 members we ship the Android app.