Inside Details of Trump Admin’s Anthropic Crackdown that Exposes Deepening AI Security Divide

The Trump administration's decision to impose sweeping export controls on Anthropic's most advanced artificial intelligence models has revealed an extraordinary clash between Washington and one of America's leading AI companies.

The development has underpinned growing concerns inside government circles that frontier AI systems may be advancing faster than existing safeguards can contain.

New details published by Politico show the restrictions were imposed only after a frantic last-minute effort by senior administration officials to persuade Anthropic to voluntarily withdraw its newly released Claude Fable 5 model, which officials believed posed significant national security risks.

According to multiple administration officials cited in reports, the dispute escalated after concerns emerged that Fable 5's security protections could potentially be bypassed, allowing users to access capabilities related to identifying software vulnerabilities and cyber weaknesses.

One of the most significant developments was the reported intervention by Amazon CEO Andy Jassy. According to people familiar with the matter, Jassy was among technology executives who raised concerns with senior Trump administration officials regarding the security implications of Anthropic's latest models.

Amazon's involvement carries particular weight because it is one of Anthropic's largest strategic investors and cloud partners. The company has committed billions of dollars to Anthropic and hosts many of its AI services through Amazon Web Services.

Reports indicate Amazon was responding to a request from administration officials for feedback on the model's capabilities and security profile.

Administration officials said findings presented by Amazon were subsequently reviewed alongside national security assessments, helping drive concerns inside the White House that the safeguards surrounding the model might not be sufficient.

An Amazon spokesperson declined to disclose details of discussions with government officials but noted that "it's not uncommon for governments to seek our counsel on potential security risks."

White House escalates concerns

The issue reportedly reached the highest levels of the Trump administration within days of Fable 5's public release. Senior officials, including Treasury Secretary Scott Bessent, Commerce Secretary Howard Lutnick, White House Cyber Director Sean Cairncross, and Chief of Staff Susie Wiles, participated in discussions about the model and possible government responses.

The administration eventually held multiple calls with Anthropic CEO Dario Amodei. During those discussions, Amodei reportedly argued that officials had misunderstood the nature of the security concerns. According to officials familiar with the calls, Amodei defended the model's safeguards and maintained that the reported vulnerability did not constitute a broad or universal jailbreak capable of disabling all protections.

Anthropic later echoed that position publicly, stating: "No testers have yet been able to find a universal jailbreak - a jailbreak method that can very broadly bypass the model's safeguards, unblocking a wide range of cyber capabilities."

The company also argued that the complete elimination of jailbreak risks remains impossible across the industry.

Anthropic said: "As we have stated publicly, we believe the government should have the ability to block unsafe deployments, as part of a statutory process that is transparent, fair, clear, and grounded in technical facts. This action does not adhere to those principles."

The White House, however, remained unconvinced. According to officials involved in the discussions, administration leaders believed Anthropic was not treating the issue with sufficient urgency.

One senior White House official said, "Export controls were a last resort after begging them for hours to work with us."

The official added: "This was not something we wanted to do, but our hands were tied."

Another person familiar with the administration's position said: "The crux of the issue was the lack of seriousness that Anthropic was applying to it."

The person added: "Had Anthropic taken it seriously and, rather than dismissing it as isolated, moved to fix or pause access, this would have never happened."

The administration ultimately imposed export controls that barred foreign nationals from accessing Anthropic's Fable 5 and Mythos 5 models. Because implementing nationality-based restrictions immediately proved operationally difficult, Anthropic responded by disabling access globally.

The development has sparked debate among AI policy experts. Even some advocates of stronger export controls questioned the breadth of the restrictions.

Jimmy Goodrich, a senior fellow at the University of California's Institute for Global Conflict and Cooperation, criticized the approach.

"This was not well thought-out," he said.

"It even bans Canadians and Brits employed at Anthropic from doing research and development."

The controversy highlights the challenge policymakers face as AI models increasingly acquire capabilities that blur the line between commercial software and technologies with potential military or cyber warfare applications.

Implications for China and Anthropic's IPO Ambitions

The restrictions are expected to have significant implications beyond the United States. Anthropic's latest models had already become a target of interest among Chinese AI developers seeking to study and replicate frontier AI techniques.

Although Anthropic's services were never officially available in China, developers frequently relied on workarounds to access the company's systems.

Industry analysts believe Fable 5's stricter controls could make that substantially harder. Kyle Chan, a fellow at the Brookings Institution, recently noted that Chinese developers may find it "nearly impossible" to use Anthropic's newest systems to accelerate development of competing models.

That outcome aligns with broader Trump administration efforts to restrict China's access to advanced AI capabilities, high-end semiconductors, and related technologies.

The timing is notable because Anthropic recently filed confidentially for a U.S. initial public offering. The company is widely regarded as one of the most valuable AI firms in the world, with an estimated valuation approaching $1 trillion.

The dispute thus introduces a new layer of regulatory uncertainty just as investors prepare to evaluate some of the largest AI-related public offerings in history. Market participants are already closely watching the recent blockbuster SpaceX IPO as a gauge of investor appetite for mega-cap technology listings.

Anthropic, OpenAI, and other frontier AI firms are expected to follow closely behind.

The latest confrontation with Washington may therefore become an important test case for how governments regulate increasingly powerful AI systems while attempting to preserve innovation and maintain national security.

A broader struggle over who controls AI

The Anthropic dispute reflects a deeper question that is increasingly shaping the AI industry: who ultimately decides how advanced AI systems can be deployed?

For years, companies such as Anthropic have argued that AI firms should help shape safety standards because they possess the deepest technical expertise. Now, the Trump administration's intervention signals a different view - that when frontier models reach capabilities with potential national security implications, government authorities may be willing to act aggressively and unilaterally.

Former White House AI adviser David Sacks summed up the administration's position after the restrictions were imposed.

"The Admin wants all of this to happen as soon as possible," Sacks wrote.

"It is frankly bewildered that Anthropic hasn't wanted to comply with safety requests that it previously said were its highest priority."