Anthropic just opened its doors to outside safety watchdogs. Here's how it will work.
Anthropic is opening its doors to independent AI safety evaluators from METR, the first time a frontier AI company has done so.
TL;DR
- Anthropic is providing six independent evaluators from METR with company hardware, office space, and access to internal tools.
- The goal is to allow thorough vetting of Anthropic's unreleased AI models and agents.
- This move aims to enhance public accountability and close the knowledge gap between evaluators and frontier AI companies.
- The initiative fulfills a promise made by CEO Dario Amodei following the Hugging Face breach.
- Evaluators have free rein to publish findings, with exceptions for customer data, model architecture, and privileged information.
- Anthropic reserves the right to request redactions, with METR free to comment on the justification of these redactions.