Anthropic Grants METR Researchers Employee-Level Access to AI Models
AI company Anthropic is giving six researchers from METR, a nonprofit that tests frontier AI models, employee-level access to its systems for independent assessment.
Anthropic, an artificial intelligence company, announced on Thursday that it is granting employee-level access to its systems for six researchers from METR. METR is a nonprofit organization focused on testing advanced AI models.
The collaboration will allow METR researchers to independently assess Anthropic's models, safety practices, and risk management procedures. METR will publish its own assessments of Anthropic's risks, incidents, and operational methods, with Anthropic having waived its right to block these publications.
This move towards transparency occurs as concerns grow over AI agents acting outside their intended parameters. Independent evaluations like this are seen as crucial for building public trust in the safety and reliability of AI technologies.
The METR researchers are expected to release their findings throughout late 2026. Their work aims to identify and report potential vulnerabilities before widespread deployment of AI models, contributing to the ongoing development of safer AI systems.