Amazon CEO Raises Security Concerns Over Advanced AI Models as U.S. Tightens Restrictions on Anthropic

BY EMMANUEL

Growing concerns over the security implications of increasingly powerful artificial intelligence systems have prompted some of the technology industry’s most influential leaders to engage directly with senior officials in the Trump administration, as Washington moves to impose new restrictions on access to cutting-edge AI models developed by Anthropic.

Among those who reportedly voiced concerns during discussions with administration officials this week was Andy Jassy, chief executive of amazon.com⁠. According to a person familiar with the matter, Jassy joined other technology leaders in warning government officials about potential security risks associated with Anthropic’s most advanced artificial intelligence systems.

The discussions come at a pivotal moment for the rapidly evolving AI sector, where governments around the world are struggling to balance innovation with concerns about national security, cybersecurity, and the potential misuse of increasingly capable machine-learning technologies.

The Trump administration took a significant step on Friday by directing Anthropic to prevent foreign nationals from accessing its newest AI models, known as Fable 5 and Mythos 5. The order applies to individuals both inside and outside the United States and was issued on national security grounds.

In response to the government directive, Anthropic announced that it would disable access to the affected models globally while complying with the new restrictions. The move represents one of the most aggressive actions taken by U.S. authorities to control access to advanced artificial intelligence systems and underscores growing concern within Washington about how such technologies could be exploited.

Anthropic said in a public statement that federal officials believe there may be a method for bypassing, or “jailbreaking,” safeguards built into Fable 5. According to the company, authorities are concerned that unauthorized users could potentially circumvent protections designed to prevent the model from being used to identify software vulnerabilities or assist in cyber-related activities.

The company indicated that the restrictions were implemented through export-control measures, placing advanced AI models under a regulatory framework traditionally used to restrict the transfer of sensitive technologies with potential military, intelligence, or strategic significance.

The development highlights the increasing view among policymakers that frontier AI systems may represent a category of technology with national security implications comparable to advanced semiconductors, cryptographic tools, and other strategically important innovations.

The U.S. Commerce Department’s Bureau of Industry and Security, the agency responsible for administering export controls, has not publicly provided detailed explanations regarding the decision. However, the action reflects growing concern within government circles that highly advanced AI systems could be leveraged by hostile actors to conduct cyber operations, accelerate military research, or gain technological advantages.

The restrictions have generated debate among experts who generally support stronger controls on advanced AI technologies. Some analysts questioned the broad scope of the order, noting that it appears to affect citizens of allied nations as well as individuals from countries considered strategic competitors.

Critics argue that applying the restrictions universally could create unintended consequences for international research collaboration and technology development. Because the order applies to foreign nationals regardless of location, it may impact researchers, engineers, and scientists working in friendly countries that maintain close security and intelligence partnerships with the United States.

Several experts have suggested that a more targeted approach focused on specific adversarial nations might have been more effective and less disruptive to ongoing scientific cooperation.

The timing of the decision is also notable because it follows indications that tensions between parts of the Trump administration and Anthropic had begun to ease in recent months. The AI company has emerged as one of the leading developers of advanced artificial intelligence systems and has attracted significant attention from investors and government officials alike.

Anthropic has confidentially filed paperwork for a U.S. initial public offering, positioning itself for a potential stock market debut that could become one of the most closely watched technology listings in recent years. The company has been competing aggressively with other major AI developers as demand for advanced generative AI tools continues to expand across industries.

The involvement of Amazon in the discussions adds another layer of significance to the issue. The technology giant has invested billions of dollars in Anthropic and has integrated the company’s AI technology into various cloud and enterprise offerings through its Amazon Web Services division. As a major strategic partner, Amazon has a direct interest in the future development and deployment of Anthropic’s models.

The broader dispute reflects a growing challenge facing governments and technology companies worldwide: determining how to regulate increasingly powerful AI systems without stifling innovation. Supporters of tighter controls argue that frontier models are rapidly acquiring capabilities that could be exploited for cyberattacks, intelligence gathering, or other harmful purposes. Opponents warn that excessive restrictions could hinder research, slow economic growth, and weaken international collaboration among democratic allies.

As artificial intelligence becomes a central arena of global competition, policymakers are expected to face mounting pressure to establish clearer rules governing who can access the most advanced systems and under what conditions. The latest action against Anthropic suggests that Washington is increasingly willing to treat cutting-edge AI technologies as strategic assets requiring government oversight, a shift that could have far-reaching consequences for the future of the global AI industry.