Anthropic suspended its Mythos 5 and Claude Fable 5 models last week following a U.S. government directive that barred foreign nationals from using the services. The company has been negotiating with the White House since Friday but has not yet reached an agreement to restore the models. The decision comes after the Trump administration deemed the models a national security risk due to concerns about their potential misuse.
Anthropic has long warned that Mythos 5 and Claude Fable 5 possess advanced capabilities for both identifying software vulnerabilities and developing exploits. The company emphasized this dual-use nature in a blog post, noting that the same queries beneficial to cybersecurity professionals and researchers could be dangerous in the hands of malicious actors. The models were initially released in a restricted form to a consortium known as Project Glasswing, with Claude Fable 5 made available to the public but with specific restrictions on biology and cybersecurity topics.
Experts argue that the regulatory response is only delaying the inevitable: other companies and open-source developers are likely to develop similar capabilities to Mythos 5. Tarah Wheeler of TPO Group stated that it is 'myopic to think no competitors will develop similar capabilities.' Logan Graham of Anthropic added that the industry must prepare for these models to become widely available within 6 to 24 months. OpenAI also released a cybersecurity-focused model in April, signaling broader industry interest in these capabilities.
Source: wired
