Anthropic Faces Dual Crises: Government Shutdown of AI Models and Consumer Lawsuit

Anthropic is fighting on two legal fronts this week. On June 12, the White House ordered the company to block foreign access to its newly released Fable 5 and Mythos 5 AI models. Days earlier, a class action lawsuit accused Anthropic of misleading users about the cost of its Claude Max subscription service.

The government directive came after Amazon researchers demonstrated a jailbreak that could make Fable 5 serve information useful in cyberattacks. Amazon CEO Andy Jassy shared those findings with the White House, according to The Wall Street Journal. Anthropic complied with the order but publicly disagreed, stating that a narrow vulnerability should not trigger a recall for a model deployed to hundreds of millions of users.

Jailbreak fears and sovereign AI debate

Anthropic removed access to both models for all customers, barring even its own foreign-born researchers from using them. The White House cited national security concerns, with reports that a China-linked group may have accessed Mythos 5. The Trump administration has not confirmed that claim. The incident has sparked a broader debate about reliance on U.S. AI, with some policymakers arguing countries should develop their own sovereign AI capabilities rather than depend on American technology subject to sudden export controls.

• The White House order blocks foreign nationals from using Fable 5 and Mythos 5, even Anthropic employees.
• Amazon's research reportedly found prompts that could extract cyberattack-related information from Fable 5.
• Anthropic denies that the models can be effectively jailbroken and disputes the government's rationale.
• The shutdown has been cited as evidence that non-U.S. AI development is necessary for technological sovereignty.
• Anthropic is in talks with the Department of Commerce and the CIA, with the government demanding unhackable LLMs.

Class action alleges deceptive pricing on Claude Max

Separately, a class action lawsuit filed in California accuses Anthropic of misleading users about the affordability of Claude Max. The suit claims that consumers signed up for a $20 per month subscription believing it offered unlimited access, only to encounter steep price increases or usage caps after minimal use. The complaint cites soaring AI costs as an existential problem for frontier models, with Anthropic allegedly failing to disclose that heavy users would face bills far exceeding the advertised rate.

Anthropic has not yet filed a response to the lawsuit. The company is also navigating a separate dispute with the Pentagon over earlier AI deployments. As both legal battles unfold, regulators and foreign governments are watching closely. The outcome could set precedents for how AI companies handle national security directives and consumer expectations around pricing.