The United States government has issued an unprecedented directive to Anthropic, compelling the artificial intelligence firm to suspend access to its most advanced AI models, Claude Fable 5 and Mythos 5, for all foreign nationals globally. This sweeping order, which includes users in India, has forced Anthropic to abruptly disable these cutting-edge models worldwide, citing national security concerns raised by US authorities.

At the heart of the US government's apprehension lies the belief that safeguards built into Fable 5 could potentially be bypassed, or “jailbroken.” This concern specifically targets the powerful cybersecurity capabilities embedded within the underlying Mythos model. Officials fear that if these advanced AI systems fall into the wrong hands, they could be exploited to identify software vulnerabilities or facilitate sophisticated cyberattacks against critical infrastructure.

Reports from The Wall Street Journal indicate that these concerns intensified following recent research by Amazon. Researchers allegedly used a series of prompts to extract information about a limited number of software vulnerabilities from an Anthropic model. These findings were reportedly shared with the US Department of Commerce, which subsequently issued the restrictive order. The directive applies universally, affecting foreign nationals both inside and outside the United States, and even extends to Anthropic's own foreign-national employees.

Anthropic, however, disputes the broad scope of the government's restriction. The company argues that the alleged “jailbreak” was narrow, limited to a handful of previously known vulnerabilities, and that other publicly available AI models can identify similar flaws without requiring any bypass. Despite its disagreement, Anthropic stated it had no alternative but to comply with the government's directive, disabling Fable 5 and Mythos 5 entirely while it works to resolve the issue and restore access.

This decision represents a significant turning point in the global regulation of artificial intelligence. Historically, government restrictions on AI have primarily focused on controlling the export of chips and hardware necessary to build and operate these systems. The US government's move against Anthropic, however, marks the first instance where a nation has directly restricted access to AI *software* itself, elevating advanced AI models to the status of critical national security assets.

The implications for countries like India and the broader international community are substantial. Zoho founder Sridhar Vembu characterized the restrictions as a “massive wake-up call” for India, asserting that “globalisation is dead.” He warned that nations can no longer assume unfettered access to the world's most advanced technologies. Vembu advocated for India to significantly increase investment in homegrown AI research and prioritize the development of sovereign AI capabilities, rather than relying solely on foreign AI companies.

This recent order also surfaces against a backdrop of existing tensions between Anthropic and various US government agencies. Earlier this year, Anthropic reportedly declined to permit its AI models for use in domestic surveillance or fully autonomous weapons systems. This disagreement is believed to have fueled broader conflicts with several government and defense officials. Furthermore, the AI firm is challenging a Pentagon decision that labeled it a “supply-chain risk,” even as it confidentially filed for an initial public offering (IPO) that could value it at nearly a trillion dollars.

For now, Anthropic remains in compliance with the US directive, with Fable 5 and Mythos 5 inaccessible to foreign users. The company continues to engage with authorities, maintaining that the government's concerns stem from a misunderstanding of a limited issue rather than a widespread security threat. The larger question, however, extends beyond Anthropic: this incident suggests that advanced AI models may soon be regulated with the same strategic importance as semiconductors, potentially compelling nations worldwide to accelerate their own domestic AI ambitions to maintain technological relevance and security.