‘Dangerous’ AI Models Are Coming No Matter What

The United States government has restricted access to Anthropic's Claude Fable 5 and Mythos 5 AI models, citing national security risks from their advanced software vulnerability exploitation capabilities, even as researchers warn that comparable capabilities will soon emerge from competitors and open-source projects [1]. Anthropic took both models offline late last week following an export-control directive that bars "any foreign national" from using the services [1]. The company has been in talks with the White House since Friday but has yet to secure an agreement to reinstate the offerings [1]. Mythos 5 was released privately to a select consortium known as Project Glasswing, while Claude Fable 5 — a Mythos-grade model — was released to the general public with blocks on responses about biology and cybersecurity [1]. The Trump administration moved to restrict both models based on its assessment that Fable 5's guardrails can be disabled, granting full access to Mythos 5 capabilities [1]. Anthropic, founded in 2021 by former OpenAI members and now valued at an estimated $965 billion as of May 2026, has positioned AI safety as a central focus [10]. The company's Claude series of large language models is trained using "constitutional AI," a technique developed to improve ethical and legal compliance [11]. Large language models are neural networks trained on vast text corpora for natural language processing tasks, and biased or inaccurate training data can make outputs less reliable [2]. Anthropic itself acknowledged the dual-use nature of the technology. "A great deal of advanced usage of AI models is dual use: the same queries that are beneficial in the hands of cybersecurity professionals and biology researchers could be dangerous if available to malicious actors," the company wrote in a blog post [1]. Security researchers argue the restriction addresses a single company rather than the broader trend. "It's myopic in the extreme to think that no other competitors to Anthropic will develop similar capabilities to Mythos or even that they have not already done so," said Tarah Wheeler, chief security officer of TPO Group [1]. Bruce Schneier, a researcher at Harvard University and the University of Toronto, said smaller, cheaper, open-source models can match Mythos and Fable's performance with more sophisticated prompting, and other models should match their capabilities "within months" [1]. The AI sector has experienced rapid growth during the 2020s, with generative AI technologies from OpenAI, Google, and Anthropic driving what is sometimes called an AI spring [3]. OpenAI, which released ChatGPT in November 2022 and was valued at $500 billion in October 2025, also conducted a private release of a cybersecurity-focused model in mid-April [4][1]. Logan Graham, Anthropic's frontier red team lead, said when Mythos Preview launched in April: "The real message is that this is not about the model or Anthropic. We need to prepare now for a world where these capabilities are broadly available in 6, 12, 24 months" [1]. Chris Wysopal, cofounder of Veracode, framed the policy question: "The policy question is not whether a technology has risk. The question is whether a specific restriction meaningfully reduces that risk or whether it mainly slows down the people trying to make systems safer" [1]. US federal agencies had already begun phasing out Claude usage after Anthropic refused to remove contractual prohibitions on mass domestic surveillance and fully autonomous weapons, leading the Department of Defense to designate the company a supply chain risk — a designation temporarily blocked by a federal judge on March 26, 2026 [11].