U.S. government forces global shutdown of Anthropic’s flagship Claude models
The U.S. government ordered AI developer Anthropic to shut down its Claude Fable 5 and Claude Mythos 5 models globally due to national security concerns, citing a potential jailbreak of Fable 5. Anthropic complied but stated disagreement with the directive, arguing that the alleged flaw is common and that such broad action would halt new model deployments across the industry. This event underscores the growing tension between AI frontier model deployment and government oversight.
Key Takeaways
- Directive received Friday June 12 at 5:21 PM ET requires model access to be cut off for all users worldwide
- Suspected 'narrow jailbreak' of Fable 5 reportedly identified security flaws in every major operating system and web browser
- Anthropic disagrees with the recall, claiming the identified flaws are already discoverable via rival models like GPT-5.5
- Shutdown includes Project Glasswing, a defensive program sharing Mythos with 50 vetted firms including Amazon and Microsoft
Why It Matters
The immediate suspension of a widely available commercial AI model for security reasons sets a volatile precedent for the entire frontier model market. By using export controls to effectively force a global recall, the government has signaled that advanced reasoning and coding capabilities are now treated as controlled munitions. For the B2B streaming and tech sectors, this introduces significant platform risk for any application built on high-capability models that could be toggled off without a formal regulatory hearing. Watch for whether the Department of Commerce issues a formal certification framework or if this remains an ad hoc emergency tool for future 'jailbreak' events.
Additional Context
The shutdown arrives during a high-stakes period for Anthropic, which filed a confidential S-1 for an initial public offering on June 1, 2026. Per Bloomberg and IG, the company is targeting a valuation of approximately $965 billion, potentially making it the first AI developer to debut near the $1 trillion threshold as early as October 2026. This financial momentum follows reports from Similarweb and Forbes in June 2026 indicating that Claude's web traffic surged 16% in May 2024, capturing 14% of all AI app downloads as it aggressively siphons market share from OpenAI's ChatGPT. Regulatory friction with the Trump administration has intensified in the weeks prior to the recall. Per Time and Axios, the Pentagon placed Anthropic on a supply chain blacklist in early 2026 after the company refused to allow its models to assist in fully autonomous weapons systems. On June 10, just two days before the recall, Anthropic CEO Dario Amodei published a policy paper comparing AI safety to FAA aircraft testing, ironically arguing that governments should have the authority to block dangerous deployments. With OpenAI's April 2026 release of GPT-5.5 and Google's Gemini 3.1 Pro already competing for enterprise dominance, the loss of Fable 5 and Mythos 5 creates a significant capability gap for Anthropic at a critical moment for its IPO valuation.
Read full article at techcrunch.com