The Trump administration lifted export restrictions on Anthropic's most powerful artificial intelligence models Tuesday evening, restoring global access after a two-week ban that had left allied governments scrambling to protect their own networks. Commerce Secretary Howard Lutnick announced the reversal in a letter to the company, clearing the way for Fable 5 and Mythos 5 to reach international customers within days—but the opaque process by which the administration reached this decision has left the tech industry deeply uncertain about how it will regulate AI going forward.
The controls, imposed on June 12, had prohibited Anthropic from providing access to either model. The restrictions triggered alarm across Europe, where allied governments had been eager to use Anthropic's products to detect and patch vulnerabilities in their computer networks before adversaries could exploit them. It also sparked friction within President Donald Trump's inner circle, which has struggled to balance the cyber risks of advanced AI against the competitive pressure to out-compete China in the sector.
The Secrecy Problem
What Anthropic actually did to satisfy government safety concerns remains unclear. Lutnick's letter required the company to proactively detect and address security risks, work with the government on protocols for future releases, and report malicious activity. Anthropic said it imposed additional technical measures to block ill-intentioned users from bypassing Fable's cyber guardrails and that the Commerce Department's Center for AI Standards and Innovation tested the model's safeguards. But the company also acknowledged that new safeguards could cause Fable to flag more benign requests—raising questions about whether safety and usability are now at odds.
Dean Ball, a former AI adviser in the Trump White House and incoming head of strategic futures at OpenAI, captured the industry's frustration bluntly. "We have no idea what Anthropic did to make the models 'safe,' what commitments Anthropic has made going forward, and whether or how any of this applies to other frontier models in the government's licensing queue," Ball said. The lack of transparency means companies and allied governments can't assess whether the government's approach to AI regulation is coherent or predictable.
A Pattern of Pressure
The Anthropic reversal came as OpenAI agreed to limit the availability of its ChatGPT-5.6 model under White House pressure, despite CEO Sam Altman's objections to how the company prefers to release products. The pattern suggests the administration is willing to intervene directly in product decisions—but without clear, public rules about when or how it will do so.
Paul Lekas, head of global public policy and government affairs at the Software & Information Industry Association, called the lifting of export controls a positive step but emphasized that "there remained a real need for a consistent process and framework for frontier model assessment." Without such a framework, companies face regulatory whiplash. Joe Hoefer, chief AI officer of Monument Advocacy, described the outcome as "more of a ceasefire than full resolution," suggesting further restrictions could return.
Glenn Gerstell, former general counsel of the National Security Agency and now a senior adviser at the Center for Strategic & International Studies, defended the decision to lift the controls. "We're not going to maintain our lead over China in cutting-edge AI simply by slapping export controls on the latest innovation," Gerstell said. He acknowledged that the government "has a lot more work to do to figure out the right way of balancing regulation for safety and national security reasons with promoting innovation in the AI sector."
Anthropic said it would restore access to Fable on Wednesday and work to reenable it on platforms including Amazon's AWS and Google Cloud as quickly as possible. The company said it would continue coordinating with the government to expand access to Mythos for select partners in the U.S. and elsewhere. White House chief of staff Susie Wiles framed the outcome as an unprecedented partnership between government and private sector, calling it a foundation of "America First."
Why This Matters:
This episode reveals how AI policy is being made without clear public rules or democratic oversight. When the government can impose and lift export controls on advanced technology with little transparency about the conditions attached, companies can't plan, allied governments can't rely on access to critical security tools, and the public has no way to assess whether national security concerns are being used as cover for industrial policy. The lack of a consistent framework means smaller companies and international partners face unpredictable restrictions while larger firms negotiate behind closed doors. Without public standards for what constitutes "safe" AI and how those standards are enforced, the government's regulatory approach risks becoming arbitrary—undermining both genuine security concerns and the rule of law that democratic governance requires.