Anthropic cut off access to its most powerful AI ever (on Donald Trump’s orders)

Anthropic cut off access to its most powerful AI ever (on Donald Trump's orders)

Anthropic

Anthropic cut off access to its most powerful AI ever (on Donald Trump's orders)

Fable 5 model is no longer available at Claude

Anthropic announced this Friday that it has suspended access to the two powerful artificial intelligence models it just launched, Fable 5 and Mythos 5, to comply with a United States national security order.

Just three days after having , the company stated, in a blog, that received a government directive which prohibits all foreign citizens, including those who work at Anthropic itself, from accessing Fable 5 and Mythos 5.

The directive, which was reportedly sent by the US Secretary of Commerce, Howard Lutnickjustifies the decision with national security concerns.

“The practical effect of this order is that we have to disable abruptly Fable 5 and Mythos 5 for all our customers, in order to ensure compliance with the directive”, stated Anthropic, which says it received the letter at 5:21 pm on Friday.

The company stated that the letter did not specifically specify what the Government’s concerns were.

However, according to its understanding, “the Government believes it was aware of a method toazer ‘jailbreak’ ao model Fable 5so that it could help in computer piracy operations”.

Fable 5, released on Tuesday, is a restricted version of Mythos 5a cutting-edge AI model that Anthropic decided not make available to the publicfaced with fears that it had unprecedented capabilities to identify vulnerabilities in software that could be exploited by hackers.

The unrestricted model, , was only made available to a limited set of companies and institutions.

A European Union, which gained access to Mythos in early June, after weeks of negotiations with the US, stated that the latest development reinforces “the Europe’s need for technological sovereignty”.

“We note Anthropic’s statement and We are assessing the situation”, these Thomas Regnierspokesperson for the European Commission, which this month outlined measures to reduce the 27-nation bloc’s dependence on the United States and Asia for key technologies, including artificial intelligence.

A Anthropic disse ter analyzed the “jailbreak” method at the center of the speculation and the hacking opportunities it exposed, but says he doesn’t believe Fable 5 will give hackers capabilities that are not already available through other public models.

The company claims that none of your experts in security tests it found a “universal jailbreak”, that is, a way to bypass the model’s safeguards against assistance to hackers.

“We disagree that the discovery of a limited potential jailbreak vulnerability must justify the withdrawal of a commercial model made available to hundreds of millions of people,” the company stated.

If this criterion were applied to the entire industrywe believe that, in practice, would stop all new model launches by all frontier model providers.”

Anthropic is andinvolved in a legal tug of war with the Trump administration for refusing to allow its technology to be used in mass surveillance and autonomous weaponswhich led the Pentagon to cut contracts with the company.

Source link