Anthropic Grants METR Researchers Access to Its Systems for AI Safety Evaluation

Anthropic has granted researchers at METR access to its systems for AI safety evaluation, allowing them to assess the risks of its models. The access granted to METR researchers is equivalent to that held by Anthropic’s risk assessment team. Anthropic has also encouraged sharing of information with experts, just like with colleagues involved in safety issues.

METR has received equipment and workspace from Anthropic. Anthropic has not paid METR for its cooperation, as METR does not accept donations from AI-related companies or their employees. Instead, it receives funding from individuals, charities, and government agencies such as the UK AI Safety Institute.

Anthropic is preparing for its expected IPO in 2026 and is seeking to reassure investors that safety issues will not hinder its listing. The company’s expected valuation is over $2 trillion.