Researchers at OpenAI, Anthropic and Google are challenging company leaders over AI policy and safety. At OpenAI, researchers and employees said they helped move the company toward closer work with state lawmakers; the company says its policy positions reflect its own assessment of its technology and how quickly its capabilities are advancing.
Researchers step into the AI policy fight
The reported dispute reaches beyond company strategy: it also touches state-level legislation and negotiations with the U.S. government. OpenAI backed an Illinois transparency bill and a California independent-auditing bill. Researchers and employees said they helped shift the company toward engagement with state lawmakers.
An OpenAI spokesperson offered a different explanation for its policy positions, pointing to the company’s technology and the pace of capability advances. Those accounts put the question of researcher influence at the center of the policy debate without making it the sole explanation for OpenAI’s decisions.
Dissent takes different forms across the industry
More than 1,000 people backed the Pacing the Frontier petition. Employees at OpenAI, Anthropic and Google also signed high-profile letters, reflecting disagreement over the direction of frontier AI companies.
At Anthropic, researcher Jacob Coxon resigned and publicly criticized the company’s safety practices. Several colleagues who remained at the company endorsed his concerns. Researchers also reportedly complicated Anthropic’s efforts to reach a deal with the Pentagon.
The criticism is not uniform in its proposed response to AI risks. Former Anthropic safety researcher Joe Benton said safety had not been prioritized to a level he considered acceptable. Sayash Kapoor, an incoming assistant professor at UC Berkeley, argued for established policy and accountability tools rather than drastic interventions.
The planned donation that did not happen
Greg and Anna Brockman’s planned second $25 million contribution to Leading the Future did not proceed. They had made a first $25 million contribution in 2025. In a June 2026 message to OpenAI employees, Greg Brockman said the couple had no plans to donate more at that time.
