The Trump administration on Friday partially lifted its ban on the export of the most advanced Anthropic spy model — easing the controversy it has sparked. caused chaos in the US AI industry.
The release clears the way for a select group of more than 100 companies and organizations to gain access to the Mythos 5 build, two weeks after the government imposed restrictions amid fears the software could be used to launch cyber attacks. But Anthropic’s second high-end model, called the Fable 5, is still under wraps.
Meanwhile, pressure from the White House caused Anthropic’s main competitor, OpenAI, to reach outreduce the release of its most advanced modelthis week because of similar internet concerns.
A series of back-and-forth moves on AI, from an administration that has promised to launch American companies to dominate the technology, has raised concerns in the industry about how President Donald Trump plans to take a heavy hand in regulating what kind of products American manufacturers can bring to market. Less than four weeks have passed since Trumpsign the executive orderrejecting mandatory federal controls on AI, instead asking leading tech companies to submit new designs to voluntary government review before making them widely available.
Talks between Anthropic and the administration are expected to continue into the weekend on restoring access to Fable, according to the person, who spoke on condition of anonymity to describe private discussions.
The talks are also focused on determining a standardized framework for evaluating cases of suspected security breaches in the future, the person said, beyond just dealing with the recent episode. POLITICO reported last weektechnical evaluation discussionwhich would form the ground rules for these types of conflicts.
“Since the release of my June 12 letter, Anthropic has worked with the US government to address the risks associated with Covered Structures,” Commerce Secretary Howard Lutnick said.he wrote in Friday’s letterto Anthropic chief information officer Tom Brown, obtained by POLITICO. “These efforts have made significant progress.”
Semaphorefirst reported on the new letterFriday night.
Anthropic, which has countered concerns about its Internet controls, issued a statement later Friday to welcome the change in management.
“We received information from the US government that Mythos 5, our most robust cyber security model, may be deployed to a small group of cyber defenders and infrastructure providers,” the company said. “We are working to provide a set of authorized service providers and restore their access to Mythos 5 as soon as possible. We are pleased to see this progress and continue to work with the government to expand access to Mythos 5 and make Fable 5 available for general use again.”
But others across the industry have expressed a desire for more clarity on the administration’s long-term intentions. Those include OpenAI, which announced Friday that, at the request of the administration, it is releasing only a short version of three versions of its new GPT-5.6 model.
“We don’t believe this type of government access process should be the long-term default,” OpenAI said in aFriday blog post. “It houses the best tools from the users, developers, businesses, network advocates, and global partners who need them.”
Friday’s changes to Mythos came after Anthropic employees and senior White House officials met over the past two weeks to discuss ways to resolve the dispute, including technical questions about whether the company has done enough to prevent its AI products from being used to launch cyberattacks. The move allows the company to restore access to its Mythos 5 models to a list of more than 100 “trusted partners,” including companies and federal agencies approved by the government, an administration official familiar with the move said on condition of anonymity to discuss it.
“In just two weeks, we’ve worked hard to ensure America remains a global leader in AI while protecting our security,” said Benno Kass, a spokesman for the Commerce Department.
The administration hadreduced export restrictionson Fable and Mythos, banning their use by foreign nationals, after a series of tense phone calls with Anthropic CEO Dario Amodei about whether the company’s security measures were as robust as promised. Anthropic then announced it was cutting access “to all our customers to ensure compliance” — while emphasizing that “we believe this is a misunderstanding.”
Amodei and Trump later sat apart from each other during a G7 lunch with allied leaders and AI executives last week in Évian-les-Bains, France.
Administration ofgreat efforts to address the issue of securityraised many questions about how it would handle future security concerns in the fast-growing AI industry, as well as whether Trump’s aides were.distinguish Anthropic unfairlyafterprevious conflicts with the company. It also left US allies in Europe and Canada suddenly unable to use Anthropic’s new models to identify and fix their cybersecurity vulnerabilities, and renewed calls for the EUreduce its dependence on American technology.
At the same time, reports ofrapid development in Chinese AI technologyhave raised concerns across the industry about whether US companies will lose their competitive advantage while awaiting administration approval.
