The United States government imposed strict export controls on Anthropic after a rapid 24-hour series of high-level discussions over its newly released AI model. The decision followed internal concerns that the company’s advanced system, known as Fable, may pose national security risks if left widely available.
According to officials, the move came after tense and urgent calls between Anthropic CEO Dario Amodei and senior members of the Trump administration. Key figures involved in the discussions included Treasury Secretary Scott Bessent and White House Cyber Director Sean Cairncross. The talks highlighted growing government concern over how fast AI technology is evolving and how difficult it is to regulate in real time.
The issue escalated quickly after the Fable model was released to the public. Within days, administration officials began questioning whether the system’s safety guardrails were strong enough. A report linked to the model raised further doubts about whether users could bypass its protections and access sensitive capabilities.
By Thursday, concerns had reached senior levels of government after Amazon CEO Andy Jassy reportedly raised security issues related to the model. The discussion involved Anthropic’s major investor Amazon, which had been asked to provide feedback to the White House on AI safety risks.
On Friday morning, top officials gathered to decide the government’s response. The meeting included Treasury Secretary Scott Bessent, White House Chief of Staff Susie Wiles, and other senior advisors. Some participated remotely while traveling. The focus was on whether immediate intervention was needed.
Officials later attempted to contact Anthropic’s leadership but were initially unable to reach CEO Dario Amodei. The company disputed claims that he was unavailable, saying he was contacted shortly after the request and participated in calls within hours. Anthropic said it also offered other senior executives for discussion during the early stages.
When direct discussions took place, Amodei joined multiple calls with senior US officials, including representatives from the Treasury, Commerce Department, and White House policy teams. The conversations focused on whether the AI model posed a broader risk or whether the issue was limited to specific security vulnerabilities.
Amodei argued that the problem identified was narrow and did not represent a full breakdown of the system’s safety controls. He said the model’s safeguards remained strong and had been tested extensively before release. He also noted that similar capabilities exist in other widely used AI systems across the industry.
Despite these arguments, officials remained concerned. They believed the reported vulnerability could still be exploited and potentially used for cyber-related attacks. Some government experts reportedly consulted national security agencies to evaluate the findings.
Following the internal review, the administration urged Anthropic to voluntarily withdraw the model while concerns were addressed. When the company did not agree to fully remove it, the government moved forward with export controls, effectively forcing restrictions on foreign access to the system.
The export controls required Anthropic to restrict access to Fable and its related model Mythos. As a result, the company disabled access for international users and some users inside the United States to comply with the order. Anthropic said the decision had to be implemented quickly and broadly.
The White House described the action as a last resort. Officials said they preferred voluntary cooperation but felt compelled to act due to the seriousness of the perceived risk. They argued that security concerns outweighed the disruption to access.
Anthropic, however, criticized the decision, saying it was disproportionate and lacked transparency. The company said it supports strong safety standards in artificial intelligence but believes regulatory actions should be based on clear technical evidence and fair procedures.
The incident has intensified debate over AI governance and national security. It also highlights growing tension between rapid technological innovation and government oversight. Policymakers are now facing pressure to define clearer rules for how advanced AI systems are released and controlled.
The White House said it remains committed to both innovation and security, while emphasizing that future AI systems must be developed with stronger safeguards to prevent misuse.
