AI Startup Faces Regulatory Storm As Washington Expands Controls Over Frontier AI Models
Artificial intelligence startup Anthropic has suspended access to its most advanced AI models for foreign nationals after receiving a directive from the US government citing national security concerns.
The company said it was ordered to disable access to Fable 5 and Mythos 5, its newest generation of frontier AI models, in what could become a landmark moment for global AI regulation. The move marks one of the first major instances of the US government restricting access to advanced AI systems directly, rather than focusing solely on the chips and infrastructure used to build them.
According to Anthropic, the directive was issued under US export control rules and required the company to halt access for foreign nationals. The startup said it was not provided with detailed technical evidence and only received verbal information regarding a potential security concern.
US Concerns Center Around AI “Jailbreak” Risks
Anthropic believes the government’s concerns relate to a potential “jailbreak” technique that could allow users to bypass safety protections built into the models.
The company said authorities fear the models could potentially be manipulated to identify software vulnerabilities, creating cybersecurity risks if exploited by malicious actors.
While Anthropic acknowledged awareness of a limited safeguard bypass, it argued that the issue does not justify withdrawing access to a commercial AI system used by millions of customers worldwide.
The company maintained that Fable 5 and Mythos 5 were launched with extensive safety guardrails and had undergone evaluation before deployment.
Relationship Between Anthropic And Washington Has Deteriorated
The latest dispute comes amid growing tensions between Anthropic and US government agencies.
Reports suggest relations worsened earlier this year after Anthropic declined to allow its AI technology to be used for domestic surveillance operations and fully autonomous weapons systems.
The disagreement reportedly resulted in the company being placed on a government supply chain blacklist scheduled to take effect later this year.
The development is particularly notable because Anthropic has frequently advocated stronger AI governance and oversight. The company has publicly supported regulations requiring advanced AI systems to undergo safety testing and risk assessments before deployment.
However, Anthropic argues that the current directive does not align with transparent and evidence-based regulation.
A New Era Of AI Export Controls
For years, US export restrictions focused primarily on limiting access to advanced semiconductors and AI computing hardware.
The Anthropic directive signals a significant shift, with regulators now targeting access to AI models themselves.
A US official confirmed that the Commerce Department ordered Anthropic to suspend access to Fable 5 and Mythos 5 for foreign nationals, potentially creating a new framework for regulating frontier AI technologies.
Industry observers believe similar measures could eventually affect other leading AI developers, including OpenAI, Google DeepMind, and xAI.
IPO Ambitions Face Fresh Uncertainty
The timing of the directive is particularly significant because Anthropic recently filed confidentially for a US initial public offering (IPO).
The company has emerged as one of OpenAI’s strongest competitors and is widely regarded as one of the world’s most valuable AI startups.
The suspension of its flagship models introduces fresh uncertainty for investors, especially as regulators worldwide increase scrutiny of advanced AI systems and their potential risks.
Amazon Web Services (AWS), a major Anthropic partner, confirmed that it was asked to revoke access to the affected models across all regions to ensure compliance with the government order.
Anthropic said it is actively engaging with authorities and believes there may have been a misunderstanding. The company added that it hopes to restore access to the models as soon as possible.
As governments move from regulating AI infrastructure to regulating AI capabilities themselves, the dispute could become a defining moment in the global debate over AI safety, national security, and access to next-generation artificial intelligence.
