The arrangement puts CEO Dario Amodei's September call for embedded oversight into practice, with publication rights and defined limits on what evaluators can see.
By [Ryan Merket](https://runtimewire.com/author/ryan-merket)
· Published
Primary source: [Anthropic](https://www.anthropic.com/news/an-update-on-independent-evaluators)
Why it matters #
Anthropic is turning a CEO-level safety pledge into an operating arrangement: six outside evaluators get internal access and publication rights, while the lab retains narrow redaction powers and controls the access tiers.
On October 15th, Anthropic began giving six employees of the nonprofit METR access to internal systems and risk-assessment work, putting CEO Dario Amodei's call for embedded oversight into practice. METR says it will publish its risk assessments, and Anthropic says it cannot veto them.…
source & further reading
runtimewire.com — original article
FTC investigates Anthropic and OpenAI over AI safety incidents
Meta releases Muse Spark Open, an 816B-parameter model built for agents
NVIDIA releases a 120B open multimodal model for AI agents