AI regulation: OpenAI and Anthropic navigate Trump administration approval process
OpenAI and Anthropic navigate Trump's ad-hoc AI model approval process, balancing government vetting demands with innovation pressure.
Objective Facts
OpenAI released its most advanced model series, GPT 5.6, to the public Thursday after delaying the public rollout at the request of the Trump administration. Anthropic announced hours later that the Trump administration has approved a limited release of its strongest cybersecurity model, two weeks after the U.S. Commerce Department effectively banned it. OpenAI said its new AI product, called GPT-5.6 Sol, would be accessible only to customers approved by the Trump administration, noting in a statement that it does not believe this kind of government access process should become the long-term default. The government and the world's most advanced AI companies are negotiating how people get access to powerful technologies case-by-case, in real time, with additional testing and meetings between the company and government officials taking place in recent weeks. Stanford University cybersecurity expert Alex Stamos said that pretty much nobody in the cybersecurity industry believes that there is any factual basis for the government's actions against Anthropic.
Left-Leaning Perspective
U.S. Rep. Lori Trahan, a Massachusetts Democrat and co-author of a bipartisan bill that would regulate AI, expressed concern that the Trump administration is deciding company by company who gets access to the newest AI model, with no law, process, or oversight, and just appointees in Washington deciding who's in and who's out. The left frames the approval process as arbitrary and lacking transparency. The apparent influence of the government on the two AI firms' delayed rollouts spurred backlash and confusion from AI policy advocates, who warned the moves signaled the White House was taking an ad hoc approach to AI regulation that could hurt innovation and set a dangerous precedent for how much influence the government can have on AI model releases. Adam Kovacevich, the CEO of Chamber of Progress, a center-left lobbying group that represents tech companies like OpenAI and Google, wrote that American AI innovation is running into Trump's patronage-and-tribute administration.
Right-Leaning Perspective
Neil Chilson, former chief technologist for the Federal Trade Commission during Trump's first term, and Adam Thierer, a senior fellow at R Street Institute, who had been very supportive of the Trump administration's less regulatory approach to AI, wrote that the latest US government action significantly escalates the centralization of control over advanced computation in the country. David Sacks, Trump's AI czar, wrote that a year ago President Trump declared America was in a global AI race and that the way to win it was to be pro-innovation, and that we deviate from that strategy at our peril. The right views the approval process as a departure from pro-innovation promises and an inconsistent shift in policy direction. Sriram Krishnan, former White House AI policy adviser, said in an interview with the Financial Times that there will not be an AI equivalent of the U.S. Food and Drug Administration, and that requiring legal reviews and administrative approval every time an AI model is released through a centralized regulatory body would undermine America's competitiveness in AI.
Deep Dive
The Trump administration entered its second term promising regulatory relief for AI companies, with officials like David Sacks explicitly opposing licensing regimes and advocating for a libertarian approach to innovation. Yet by June 2026, the same administration began issuing export controls on Anthropic's models and requesting staggered releases from OpenAI—moves that look functionally like the ad-hoc vetting it opposed. The trigger was Anthropic's public warning that its Mythos model could be weaponized for cyberattacks, allegedly flagged by Amazon CEO Andy Jassy to the administration. This shifted the White House calculus toward treating frontier AI as a national security asset requiring case-by-case approval. The right's critique—from Trump loyalists like Dean Ball who helped draft the original AI action plan—identifies a genuine contradiction: the administration set a clear pro-innovation policy direction, then undermined it through discretionary executive action. However, the right stops short of calling this a fundamental problem with the approach itself; instead, it characterizes the moves as a policy implementation error. The left, meanwhile, sees the approval process as revealing the true nature of Trump's AI stance: not principled deregulation, but selective favoritism. Critics point out that OpenAI, whose leadership donated heavily to Trump and praised him publicly, faced requests rather than binding orders, while Anthropic—whose CEO called Trump a "feudal warlord" and opposes Trump-friendly deregulation—faced export controls. The asymmetry suggests political motivation, though both companies now navigate the same approval framework. What remains unresolved is whether the approval process will crystallize into transparent rules or remain a tool of ad-hoc executive discretion. The administration claims the vetting is temporary and that Trump's June 2 executive order establishing a voluntary 30-day review framework will provide structure. Yet the case-by-case negotiations with OpenAI and Anthropic, ongoing as of early July, suggest the process remains opaque. Neither company has named which government-approved customers access their models. Neither side fully addresses what happens when the next model poses ambiguous security tradeoffs that experts disagree about—a likely scenario given the novelty of frontier AI systems. The stakes are high: this approval process sets a precedent for how much discretionary power the executive branch claims over private AI development, with implications for competitiveness against China (a concern both sides raise) and for the future of private tech innovation.