US Blogger Accuses Anthropic of Influencing AI Safety Evaluator via Funding Network

On September 15, US blogger Kevin Bass published an investigative report accusing Anthropic of potentially using a funding network to influence AI safety evaluation organization METR and promote AI doomsday narratives, calling on the US Congress to investigate.

The report said Anthropic had previously suggested that AI research institutions undergo METR evaluations, but the investigation found that METR was not an objective independent third party and had strong financial ties with Anthropic. It involves Facebook co-founder Dustin Moskovitz, whose Anthropic shares were placed into the Good Ventures Foundation, a major funder of the related ecosystem. The value of these shares has grown from approximately $500 million in early 2025 to $7.7 billion as of September 2026.

Kevin Bass also found that organizations with financial links to METR funded Tarbell Center and other institutions promoting AI risk and AI doomsday narratives. Related content has appeared in outlets such as The Verge, Science, Los Angeles Times, The Dispatch and TIME.

Bass argued that such funding structures could undermine the independence of assessments of Anthropic models, creating a feedback loop in which Anthropic grows, AI risk narratives expand, regulatory demand rises, and related organizations receive more funding.