Anthropic grants AI watchdog "employee-like" access

AI company Anthropic has provided six employees from the watchdog organization METR with "employee-like" access to its internal tools and systems, including company hardware and office space. This initiative aims to enhance the transparency and rigor of AI safety evaluations. Anthropic stated that it encouraged its employees to share information with the evaluators as they would with a fellow risk assessment team member. METR, which is funded by individuals, philanthropic organizations, and government bodies like the UK's AI Security Institute, does not accept donations from AI companies. The evaluators will publish their findings, though Anthropic can request redactions for sensitive information such as customer data or proprietary model architectures. Anthropic is also collaborating with Accenture on safety testing, with both companies committing at least $1 billion to establish new safeguards. Another third-party tester, SecureBio, reviews chemical and biological weapon risks for Anthropic.