Anthropic Grants METR Researchers Full Employee-Level Access
Anthropic has removed its ability to block METR's evaluations, giving the research team employee-level access to examine the company's AI risks and incidents.
Anthropic, a leading frontier AI model developer, has taken a notable step toward transparency by granting the Machine Ethics and Trust Research (METR) team unrestricted employee-level access to its systems. This access enables METR to conduct independent assessments of Anthropic’s risk management, incident handling, and overall operational practices.
The arrangement explicitly removes Anthropic’s right to block or limit METR’s investigations, signaling a willingness to expose internal processes to external scrutiny. METR plans to publish its findings, offering the broader AI community insight into the safety and governance measures employed by the company.
By opening its internal environment to METR, Anthropic aims to foster trust and demonstrate accountability in the rapidly evolving AI landscape. The move aligns with growing calls for greater openness around AI development and risk mitigation strategies.
TechCrunch AI reported the development on October 15, 2026, highlighting the significance of the partnership for both research transparency and industry standards.
The collaboration may set a precedent for other AI firms seeking to balance proprietary concerns with the demand for independent safety evaluations.
Read also
Anthropic Halts Live Internet Access for Internal AI Evaluations
Anthropic has turned off live internet access for its internal AI evaluations, pending further notice.
Anthropic Launches Claude Haiku 5.5 with Lower Pricing
Anthropic's new Claude Haiku 5.5 offers faster performance and a price point that aligns with OpenAI's GPT‑6 Luna for the first 100,000 tokens.
Max Hodak Unveils PRIMA Progress at TechCrunch Disrupt
At TechCrunch Disrupt, Max Hodak detailed PRIMA's role in Science Corporation's journey toward a novel BCI.
