Anthropic gives METR employee-like access to evaluate its models and safety practices
Article timeline
The articles in this story. Times are ET.
- TechCrunch · AINewsAnthropic gives METR researchers employee-level access to assess its models
AIAnthropic is giving six researchers from the nonprofit METR employee-level access to independently assess its models, safety practices, and risk management. METR will publish its own assessments and Anthropic cannot veto publication, though it can require redactions in narrow categories such as customer data and model architectures. METR is currently the only evaluator in the top tier of a three-tier system Anthropic is building.
Rohan Paul@rohanpaul_aiXPickAnthropic gives six METR safety evaluators employee-like access to internal systemsAIAnthropic has onboarded six METR employees with company hardware, office space, and internal tools and systems roughly on par with its own risk assessment teams. METR can publish its own verdicts on Anthropic's models and practices, and Anthropic can only require redactions in narrow, predefined categories such as customer data. Other evaluators, such as SecureBio, start at the lowest of three access tiers, where Anthropic still controls what they see and approves what they publish.
Andrew Curran@AndrewCurran_XMETR gets employee-level access to Anthropic's frontier modelsAIMETR now has employee-level access to Anthropic's frontier models, according to Andrew Curran. Anthropic says it has committed to having evaluators assess its safety practices and has given a METR team employee-like access, while discussions with other organizations continue.
Anthropic@AnthropicAIOfficialAnthropic gives METR employee-like access to assess its safety practicesAIAnthropic says it has committed to having evaluators assess its safety practices and is in discussions with multiple organizations. It has now given employee-like access to a team from METR, and a linked post describes how it works with third-party evaluators.
- Anthropic NewsroomOfficialPickAnthropic gives METR employee-like access to evaluate its frontier models
AIAnthropic says it has onboarded six METR employees with company hardware, office space, and internal access roughly comparable to its own risk assessment teams. METR will publish its analysis of Anthropic's risk levels, incidents, or practices, and Anthropic can require redaction only of narrow categories such as customer data, privileged information, and model architectures. Anthropic also describes three access tiers for other evaluators, including SecureBio, which is currently in the first tier reviewing its chemical and biological weapons chapter.