Trump authorizes Anthropic to (partially) open its most powerful AI outside the US

The decision allows a restricted group of companies and agencies to once again have access to the company’s Mythos 5 model. A second advanced Anthropic model, however, remains blocked.

The Trump administration partially lifted, this Friday, the applied to one of Anthropic’s most advanced artificial intelligence models, in a retreat that seeks to end two weeks of tension and uncertainty in the North American AI industry.

The decision paves the way for access to Mythos 5 by a restricted group of more than 100 companies and agenciestwo weeks after the administration imposed limitations over fears the software could be used to launch cyberattacks.

One second Anthropic’s advanced model, the Fable 5continues, however, blocked.

At the same time, pressure from the White House led to OpenAI, main competitor from Anthropic, a restrict this week the launch of your model more advanced, due to similar cybersecurity concerns.

The succession of advances and setbacks in an area where the administration has promised to give North American companies room to lead in technology is raising concerns in the industry. At issue is the degree of interventionwhat President Donald Trump intends to exert on the products that US programmers can put on the market.

Less than four weeks ago, Trump signed an executive order rejecting mandatory federal controls on AI and asked, alternatively, the main technology companies tosubject new models to an evaluation government’s voluntary approach before making them available on a large scale.

Talks between Anthropic and management are expected to continue over the weekend, with the aim of restore access to Fableaccording to a source interviewed, on condition of anonymity, by .

Discussions have focused on creating a standardized frame to evaluate future cases of suspected contour of security mechanismssaid the same source.

Friday’s reversal on Mythos came after Anthropic officials and senior White House officials met over the past two weeks to try to resolve the impasse. The conversations included technical questions about whether the company had done enough to prevent your AI products from being used in cyber attacks.

The move allows Anthropic to restore access to Mythos 5 models to a roster of more than 100 “trusted partners”including companies and federal agencies approved by the Government, said an administration official familiar with the decision, who was granted anonymity to discuss it.

“In just two weeks, we have worked diligently to ensure that the United States continues to be a world leader in AI without compromising our security,” he said Benno the Catspokesperson for the Department of Commerce.

The administration had imposed restrictions on the export of Fable and Mythos, prohibiting its use by foreign citizensafter a series of tense calls with Dario Amodei, CEO of Anthropic, about the strength of the company’s security mechanisms.

Anthropic then announced that it would “cut off access to all our customersto ensure compliance” with the rules, insisting at the same time that “we believe this is a misunderstanding”.

The decision left U.S. allies in Europe and Canada suddenly unable to use Anthropic’s new models to identify and fix vulnerabilities cybersecurity in your own systemsrenewing calls for the EU to reduce dependence on North American technology.

However, reports about China’s rapid advances in AI increased concern in the sector about the risk of North American companies losing competitive advantage pending approvals from the administration.