Anthropic Embeds Independent Safety Team to Monitor Claude Models
Anthropic has opened its lab to a nonprofit watchdog, granting deep access to its models and internal systems as part of a broader effort to curb potential runaway AI.
Anthropic has opened its production environment to a nonprofit evaluation team, granting them “employee‑like” presence to inspect the company’s Claude models in real time.
The nonprofit, METR, now operates within the lab and can examine hardware and internal systems, providing a fresh layer of oversight into how Claude behaves.
This arrangement follows a broader industry move to involve third‑party auditors amid mounting concerns that as AI capabilities grow, unchecked releases could threaten the economy, cyber security and society more broadly.
Sources1
Original reporting and primary evidence.
Claim connections3
Anthropic has granted external evaluators 'employee-like access' to assess the risks of its models, following through on a commitment to increase safeguards over concerns the technology could run catastrophically out of control.
confirmedConnected sources1
METR, a nonprofit that has worked with leading labs including OpenAI, Anthropic, Google and Meta, now has a team embedded in the Claude maker's office with access to its hardware and internal systems.
confirmedConnected sources1
A greater role for third-party evaluators is among the main concrete steps the labs have proposed in response to rising concern about the dangers more capable AI systems could pose to the economy, cyber security and humanity at large.
confirmed