APREX · EST. 2026
On Friday evening, AI company Anthropic found itself at the center of a political storm when the U.S. government issued a direct order: its two newest models, Fable 5 and Mythos 5, were to be made immediately inaccessible to all foreign actors. The order applied not only to customers and partners — it also covered Anthropic's own foreign-born employees.
Shut down to ensure compliance
To comply with the order, Anthropic chose the simplest — and most drastic — solution: switching off access to both models for all customers, regardless of nationality or affiliation. The company justified this by saying it was the only practically feasible way to ensure full compliance on short notice, according to The Verge.
In a public statement, Anthropic confirmed that it was complying with the order, but noted that the authorities "did not provide specific details about the national security concern." The underlying rationale is reportedly rooted in fears that the models could be "jailbroken" — meaning that built-in safety mechanisms could be circumvented by actors with malicious intent.
The government provided no specific details about the national security concern — Anthropic in public statement
A policy without precedent
Experts are characterizing this as an unusual and far-reaching move. Chris McGuire, senior researcher on China and emerging technologies at the Council on Foreign Relations, is quoted in background materials with a clear assessment: targeted access controls can be justified, but a total shutdown of all countries without any prior warning is "highly questionable."
The move represents an escalation in U.S. AI policy, which according to research materials has shifted away from a primary focus on ethics and safety and toward competitiveness and geopolitical dominance. "America's AI Action Plan," launched in July 2025, points clearly in this direction.
Europe responds with calls for independence
Reactions across Europe were swift. A spokesperson for the European Commission reportedly stated, according to research materials, that the incident "further underscores Europe's need for technological sovereignty." This is not the first time warnings have been raised about becoming too dependent on American AI platforms, but a concrete example of access being cut without notice lends the argument new weight.
U.S. policy also risks proving counterproductive. Rowan Wilkinson of Chatham House argued in April 2026 that hardware export controls alone will not stop China's AI development, due to loopholes, smuggling, and algorithmic advances that have allowed China to move beyond a hardware-dependent approach. Daniel Castro points out that Nvidia restrictions could accelerate China's technological self-sufficiency — and thereby weaken long-term American influence.
A fragmented regulatory landscape
The incident comes at a time when the absence of comprehensive federal AI legislation in the United States has created a confusing regulatory environment. According to a survey from February 2026, 58 percent of European and British developers are experiencing delays due to regulation, and more than a third have been forced to remove or downgrade features in order to meet compliance requirements.
What Anthropic will now do to restore parts of the service — potentially through technical solutions that satisfy the government's requirements — remains unknown. The company has not published any timeline for when Fable 5 and Mythos 5 might become available again.