Trump Releases Mythos and Fable: Anthropic's AI Now Accessible Again

The United States Lifts Restrictions on Anthropic's AI Models

U.S. authorities have decided to lift the requirement that forced Anthropic to obtain a license before exporting its artificial intelligence models, Mythos and Fable, internationally. This requirement had effectively restricted public access to these models, which are considered among the most advanced ever developed in the field of AI.

Anthropic announced that starting Wednesday, July 1, access to these models would be gradually restored. This decision marks a turning point after, on June 12, the U.S. government had placed these products on a restricted export technology list. This designation meant that they could only be accessible to foreign nationals with special authorization, a constraint that was difficult to manage on a large scale, prompting Anthropic to suspend public access.

A Compromise with the U.S. Government

After several weeks of negotiations, Secretary of Commerce Howard Lutnick announced that Anthropic had committed to proactively detecting and addressing security risks associated with its models. The company will work closely with the U.S. government to establish protocols, standards, and publications for Mythos, Fable, and future models. Additionally, Anthropic has pledged to inform the government of any detected malicious activity.

Anthropic had already taken the initiative to implement much of these measures voluntarily, well before the introduction of the export rule. This proactivity had led some cybersecurity experts to doubt the effectiveness of the initial restrictions, viewing them more as a lever to punish Anthropic for the public criticisms of its leaders.

International Pressures and Asian Competition

Initially, Mythos had been made available to a select group of organizations as early as April, to alleviate concerns about its ability to identify and exploit vulnerabilities in software. A version named Fable was made public in June, with enhanced security measures.

However, in the face of the rising power of Asian AI companies, which are beginning to offer models with capabilities close to Mythos, such as Fugu and Tulongfeng, the U.S. government was compelled to reassess its position. The goal is to ensure that American AI maintains its competitiveness on a global scale.

An Uncertain AI Policy

Last week, Lutnick gave the green light for Mythos to be accessible to selected clients approved by the White House. Similarly, the latest models from OpenAI have been made available to a group of organizations validated by the Trump administration, rather than being opened to the public.

This fluctuating approach to AI policy by the Trump administration has left companies in the sector uncertain about the rules that will govern future model releases. An executive order issued in June, which indicated a willingness to review models before their release, was criticized by Dean W. Ball.