What happened
Two organizations, the Tarbell Center and Model Evaluation and Threat Research, have been issuing warnings about the potential dangers of artificial intelligence. However, it has been found that these groups have ties to allies of the technology firm Anthropic, including Meta cofounder Dustin Moskovitz and Skype cofounder Jaan Tallinn.
The Tarbell Center has a fellowship program that places tech reporters at news outlets to publish coverage of AI, while Model Evaluation and Threat Research has been promoted by Anthropic CEO Dario Amodei as a third-party evaluator of AI safety. Both organizations have connections to Anthropic through their funders and founders.
Moskovitz and Tallinn were early investors in Anthropic, and AI researcher Paul Christiano, who is connected to both organizations, was an initial trustee for Anthropic's Long-Term Benefit Trust. Christiano has since stepped down to join the Anthropic-backed AI Safety Institute.
Critics have raised concerns that Anthropic's inner circle is trying to monopolize the public policy discussion on AI to advocate for its preferred restrictions on the technology. The company did not respond to requests for comment.
Model Evaluation and Threat Research has adopted a policy of not accepting donations from frontier AI companies or their employees, and its president, Chris Painter, has stated that the organization's goal is to bring information from inside these companies into the public domain. The Tarbell Center also claims to maintain editorial independence and does not take direct donations from companies like Anthropic.
