Anthropic Faces Export Controls, Billing Backlash, and Sovereign AI Pressure

Anthropic is under pressure from multiple fronts in June 2026 as the US government restricts foreign access to its latest AI models, cybersecurity executives push back, and the company reverses a planned billing change. The San Francisco based AI firm also hired a Meta veteran for its data center supply chain team while competitors argue the export curbs strengthen the case for sovereign AI.

More than 100 cybersecurity executives and experts from companies including Adobe and Nvidia signed a letter to the Trump administration on June 14, 2026, urging the government to lift export control directives on Anthropic's Fable 5 and Mythos 5 models. The letter argues that restricting access to these models harms US cyber defense capabilities while adversaries like China are only months behind in AI development.

Export Controls Spark Industry Backlash

Anthropic said on June 12 that it took Fable 5 and Mythos 5 offline to comply with a Commerce Department directive preventing foreign nationals from using the models. The company stated it did not believe the government's security concerns warranted the action. Anthropic had previously limited access to Mythos 5 due to its ability to surpass human cybersecurity experts in finding and exploiting vulnerabilities.

• The letter calls for an open, scientific, and transparent process for AI risk assessments in the future.
• Signatories note that while Mythos models are good at finding software flaws, they are not uniquely good, and other open source models are used for similar tasks.
• The directive came 10 days after President Trump signed an executive order establishing a voluntary framework for vetting AI systems for national security risks up to a month before public release.
• Tensions between Anthropic and the administration escalated after a contract dispute with the Pentagon over use of AI in autonomous weapons and surveillance.

Operational and Competitive Pressures

Anthropic also reversed a planned billing overhaul for its Claude Agent SDK just before launch. Instead of introducing separate credits, the SDK and third party apps will continue drawing from regular subscription limits. The move comes as a price war with OpenAI looms, according to The Decoder. Separately, Anthropic hired Sham Parmar from Meta to join its data center supply chain team, focusing on power and cooling infrastructure for buildouts, Data Center Dynamics reported.

The export controls have also fueled arguments for sovereign AI. Sung Kim, CEO of South Korean startup Upstage, said the US government's order to cut foreign access to Anthropic's models, and the company's decision to switch them off worldwide rather than comply selectively, provides evidence for countries to develop their own AI capabilities. Kim made the comments at a conference covered by The Next Web.

What comes next remains uncertain. The Commerce Department has not responded to requests for comment. Anthropic is challenging the Pentagon's attempt to declare it a supply chain risk in federal court. The company continues to navigate a landscape where its safety conscious approach clashes with both government demands and market competition.