OpenAI and Anthropic Models Faced Trump-Era U.S. Vetting Ahead of Release

According to Axios, OpenAI and Anthropic’s latest, most powerful AI models received U.S. government sign-offs before broad release under the Trump administration’s approach to AI regulation, which has included export-control threats, potential licensing requirements, and negotiations across multiple agencies. Axios reported that President Trump scrapped reporting requirements from a Biden-era AI executive order that had required companies to share safety testing results with the government, and that Amazon flagged a “jailbreaking” vulnerability last month that ultimately led to export controls on Anthropic. The White House said Tuesday it did not approve or disapprove OpenAI’s decision to release a model, while spokesperson Liz Huston said top technology leaders are backing the president’s approach; OpenAI CEO Sam Altman told CNBC the day after OpenAI said it would widely release GPT 5.6 that the government process was productive. Axios also reported that CAISI’s operational budget is $15 million but it needs $84 million annually to fulfill Trump’s AI action plan, according to the Institute for Progress, and that a voluntary framework required by a June AI executive order is due Aug. 1.