Anthropic gives outside AI watchdog “employee-like” access

What happened: Anthropic granted external evaluators "employee-like access" to assess risks of its models. METR, a nonprofit that has worked with OpenAI, Anthropic, Google and Meta, now has a team embedded in Why it matters: The move follows through on a commitment to increase safeguards over concerns the technology could run catastrophically out of control. A greater role for third-party evaluators is among the main concrete steps labs have proposed in response to rising concern about dangers more capable AI systems could pose to the economy, cyber security and humanity at large.

Anthropic granted external evaluators "employee-like access" to assess risks of its models. METR, a nonprofit that has worked with OpenAI, Anthropic, Google and Meta, now has a team embedded in

The move follows through on a commitment to increase safeguards over concerns the technology could run catastrophically out of control. A greater role for third-party evaluators is among the main concrete steps labs have proposed in response to rising concern about dangers more capable AI systems could pose to the economy, cyber security and humanity at large.