US government orders Anthropic to remove two AI models after reported concerns

US government authorities have ordered Anthropic to pull two of its AI models, identified in reports as Claude Fable and Mythos, in a move that adds another layer of scrutiny to the fast-moving artificial intelligence sector. The decision, reported by Decrypt and later echoed in broader coverage, appears to stem from concerns over a vulnerability linked to the models. Anthropic has pushed back against the demand, arguing that the weakness cited by officials is not unique to its systems and is already widespread across the industry.

The reported order places Anthropic in a sensitive position at a time when large AI developers are facing growing pressure from regulators, customers and business partners to demonstrate that their systems are secure and reliable. While the sources do not provide a detailed description of the vulnerability itself, they indicate that the issue was serious enough for the government to seek removal of the models. Anthropic’s response suggests that the company views the matter as a broader industry problem rather than a flaw limited to its own products.

According to the reporting, the dispute is not unfolding in isolation. Investing.com reported that Amazon had voiced concerns about Anthropic’s AI models before the government crackdown, citing a source familiar with the matter. That detail suggests the models were already under scrutiny from important commercial stakeholders before the government action became public. The sources do not say what form Amazon’s concerns took, but the reporting indicates that questions about the models had been circulating prior to the official order.

Anthropic’s pushback is notable because it frames the government’s action as potentially disproportionate. The company said the vulnerability referenced by officials is already common throughout the AI sector, implying that any enforcement step focused only on Anthropic could be seen as inconsistent or overly broad. The reports do not include a formal regulatory statement or the government’s full reasoning, leaving the exact grounds for the order unclear. Still, the exchange highlights a recurring issue in AI oversight: whether regulators should target specific products when problems may affect a wider class of systems.

The episode also underscores the increasingly intertwined relationship between major technology firms and AI developers. Anthropic has been one of the more prominent names in the generative AI space, and attention from both government authorities and a major corporate backer such as Amazon can have implications beyond the immediate models in question. Even without a full technical explanation in the available reporting, the fact that both official and commercial concerns emerged around the same models points to a heightened focus on model safety, security and deployment standards.

For the wider AI market, the situation serves as another reminder that rapid product development is taking place alongside rising scrutiny. Developers are being asked not only to improve capability, but also to address vulnerabilities that could affect users, customers and infrastructure. The sources do not indicate whether Anthropic intends to comply immediately, seek a compromise or challenge the order further. They do, however, show that the dispute is now public and may draw broader attention to how regulators and industry participants handle security concerns in advanced AI systems.

At this stage, the available reporting leaves several key questions unanswered, including which authority issued the order, what specific vulnerability was cited, and whether the models have already been removed or suspended. What is clear from the sources is that Anthropic is contesting the basis for the action, Amazon had already raised concerns privately according to one report, and the episode has become part of the broader debate over how AI models should be governed as they move into wider use.