Anthropic has granted external evaluators “employee-like access” to assess the risks of its AI models, fulfilling a commitment to increase safeguards against technology running out of control. The move marks a significant shift toward transparency as concerns grow regarding the impact of advanced AI on global security and the economy.
Embedded Oversight by METR
METR, a nonprofit organization that has previously worked with OpenAI, Google, and Meta, has embedded a team of six employees within Anthropic’s offices. These evaluators have been provided with company hardware, office space, and access to internal tools and systems roughly equivalent to those used by Anthropic’s own risk assessment teams.
Crucially, METR is authorized to publish its findings without a veto from Anthropic. While the company can require redactions for sensitive categories—such as customer data, legal privilege, or specific model architectures—the core analysis remains independent. Notably, Anthropic is not paying METR for this service; the nonprofit is funded by independent donations and government bodies like the UK’s AI Security Institute.
The Road to a $2 Trillion IPO
The announcement comes as Anthropic prepares for an initial public offering expected this year, with a projected valuation exceeding $2 trillion. The company has recently focused investor presentations on its safety practices to reassure the market that risk management will not overshadow growth. In addition to METR, Anthropic is partnering with Accenture, with both firms committing at least $1 billion each to develop new safeguards.
Industry Alignment and Skepticism
The proposal to embed external auditors was previously supported by industry rivals Sam Altman of OpenAI and Elon Musk. However, critics argue that voluntary arrangements may lead to conflicts of interest similar to those seen in the accounting sector, suggesting that government regulation remains necessary. Anthropic defended the model by comparing it to bank supervision, where resident examiners work inside major financial institutions to identify risks before they threaten the public.
