Trump admin in-fighting stalls AI regulation over voluntary model sharing

The Trump administration is locked in an internal struggle over whether to revive an executive order on AI regulation that President Trump abruptly canceled on May 21, just hours before a planned signing ceremony. The order would have created a voluntary framework requiring leading AI labs—including OpenAI, Anthropic, and Google—to provide the White House early access to frontier models like GPT-5.5 and Anthropic’s Mythos, which are increasingly capable of finding vulnerabilities in legacy software. The key provision mandated model submission up to 90 days before public release, though several AI executives have privately told WIRED their companies may not be ready to share models that far in advance. The push reflects a growing recognition inside the White House that AI is a fast-emerging national security concern, a shift from earlier anti-regulation stances.

White House chief of staff Susie Wiles has taken charge of a pro-regulation faction that includes Treasury Secretary Scott Bessent and national cyber director Sean Cairncross. Bessent has met with Anthropic CEO Dario Amodei and other AI leaders to forge a path forward, and is expected to lead cross-border AI regulation talks with China. Opposing them is former AI czar David Sacks, who successfully convinced Trump to cancel the signing ceremony. Sacks argues the order would stifle innovation and hamper the US race against China. Commerce Secretary Howard Lutnick remains distant from the process, having already gained early model access through the preexisting Center for AI Standards and Innovation. Ultimately, aides say the biggest hurdle is Trump himself—internal reconciliation means nothing if the president won't sign. A White House spokesperson insists the team is united in executing Trump’s agenda and maintaining the right balance.