The Trump administration's pro-AI coalition is fracturing over a fundamental tension: whether to prioritize national security or keep American companies racing ahead of Chinese competitors. The battle is unfolding in real time and could determine how the government manages what may be the most consequential technology of the decade.
David Sacks, Trump's former AI and crypto czar, fired the opening shot. He warned that restricting access to cutting-edge AI models directly contradicts the competitive strategy Trump announced just a year ago. "A year ago, President Trump declared that America was in a global AI race and that the way to win it was to be pro-innovation," Sacks posted on X. "President Trump was exactly right. We deviate from that strategy at our peril."
His complaint centers on White House orders to delay major model releases. OpenAI was asked to stagger the rollout of GPT-5.6, breaking it into phases. Anthropic faced directives to halt access to its Fable 5 and Mythos 5 models, though Mythos has since returned in limited form after Commerce Secretary Howard Lutnick noted the company's government collaboration had "yielded significant progress." Fable 5 may follow.
The stakes for the industry are enormous. Kevin Bankston, an AI governance advisor at the Center for Democracy and Technology, argued these restrictions amount to self-sabotage: "This is how you crash the U.S. AI market." Aaron Levie, CEO of Box, called the shift "one of the most important changes in the AI landscape in the past four years." He noted that constant competitive pressure between labs has been the engine driving rapid AI progress.
But there's a catch. While U.S. companies face government-imposed release schedules, Chinese rivals operate under no such constraints. Recent security evaluations show Chinese AI systems have already matched the capabilities of top American models in cybersecurity tasks. Meanwhile, open-source Chinese models are surging in adoption as developers seek cheaper alternatives, with several now ranking in the top positions on popular usage leaderboards.
Venture capitalists are sounding alarms. Paul Kedrosky, a prominent VC, told Axios the situation is "hugely bearish" for AI investors. "The AI party now has a hall monitor who is also diluting the punch," he said. The implication is clear: if the government throttles deployment of AI labs' most valuable products, investor valuations could tank.
Yet the business world isn't unified in opposing oversight. Some labs have actually requested clearer federal rules. Anthropic has pushed for stronger safeguards as models grow more capable. Dan Shipper, CEO of the AI subscription service Every, acknowledged that government involvement is necessary, but "they just need to find the right balance between safety and broad access."
What executives and investors agree on is this: ad hoc government decisions are worse than clear rules. Mark Pincus, who founded Zynga and invests in both OpenAI and Anthropic, supports regulation in principle but says "it's hard to build when there's a moving target." Benchmarks matter. Siméon Campos, an AI startup founder, warned that without clear metrics, companies could simply optimize their way around whatever rules emerge.
The real problem is opacity. Frontier AI access has become too valuable to leave to arbitrary government discretion, especially when companies don't know what the playing field will look like by the time rules are finalized.
Author James Rodriguez: "The Trump team wants to win the AI race without losing it to national security hawks, but you can't have both at once. Someone's going to lose."
Comments