Amazon CEO Andy Jassy and executives from five other tech companies warned the Trump administration about security vulnerabilities in Anthropic's Fable model. Within hours, the White House issued an export control order forcing the model offline. The timing raises questions about whether this was a security decision, a competitive move, or both.

Anthropic built Fable as a specialized reasoning model designed to compete with advanced systems from rivals like OpenAI and Google. The model reportedly showed promise in internal testing, but Amazon and the other companies flagged unspecified security issues to government officials. The White House moved quickly, treating the concerns as a national security matter serious enough to warrant immediate action.

The situation carries obvious complications. Amazon is one of Anthropic's largest investors, making the warning feel like a backstab from within the tent. Tech executives often flag competitor vulnerabilities to regulators, but the speed of the government response suggests either genuine alarm or willingness to use security concerns as a lever against a company the administration views as inconvenient.

Anthropic has not publicly disclosed what vulnerabilities existed in Fable or whether the model actually posed risks. The company also has not confirmed the details about which companies made the warnings. This lack of transparency fuels skepticism about whether the crackdown reflects real danger or political pressure.

The decision fits a pattern in the Trump administration's approach to AI: aggressive use of export controls and regulatory authority to shape which companies advance. Anthropic competes directly with OpenAI, which has closer relationships with key administration figures. This move keeps Anthropic's more capable models from reaching international users and potentially damages its credibility with enterprise customers.

Whether Fable had genuine flaws or not remains unclear. What is clear is that security concerns now serve as a useful tool for controlling which AI systems reach the market.