Anthropic halted access to its latest AI models, Claude Fable 5 and Mythos 5, after U.S. national security authorities ordered the company to block foreign nationals from using them. The move came just days after the public launch. In its statement, Anthropic said regulators flagged a way to “jailbreak” Fable 5, bypassing built-in restrictions. Although the company found only a handful of minor, previously known flaws—ones other public models could also expose—it disabled both versions to stay in compliance. Before the suspension, Anthropic had touted multiple safeguards against hacking. It gave a small group of organizations preview access in April to test for vulnerabilities. Critics suggested the “too powerful” marketing line overstated the model’s hype. Yet the AI indeed matched or outperformed its rivals, including OpenAI’s ChatGPT and Google’s Gemini, in early benchmarks. In Brussels, the European Commission said the suspension highlights Europe’s need for technological independence. A spokesman, Thomas Regnier, noted the bloc is reassessing its AI plans after unveiling steps to cut reliance on U.S. and Asian tech. Back in London, Queen Mary University’s Gina Neff warned that restricting access could hamper both secure testing and government collaboration. She pointed to U.K. tests showing the model could breach defenses 73 percent of the time. Anthropic is also locked in a lawsuit with the Trump administration over a Pentagon move labeling it a “supply chain risk.” That designation bars government agencies and military partners from using its tools—traditionally reserved for firms in rival nations. A federal judge has blocked enforcement of the Pentagon’s order, so agencies can continue using Anthropic’s services while the case proceeds.