Amazon CEO Raises AI Security Concerns as U.S. Orders Anthropic to Restrict Advanced Models Worldwide

BY COMFORT OGBONNA

Amazon Chief Executive Officer Andy Jassy was among a group of technology leaders who reportedly expressed concerns to senior officials in the Trump administration this week regarding the security implications of Anthropic’s most advanced artificial intelligence models. The discussions highlight growing concerns within both the technology industry and government circles about the potential risks posed by increasingly powerful AI systems.

The concerns emerged shortly before Anthropic took the extraordinary step of disabling access to its newest AI models worldwide following national security directives issued by the U.S. government. The move marks one of the most significant government interventions in the rapidly evolving artificial intelligence sector and underscores the growing tension between technological innovation and national security considerations.

Anthropic, a San Francisco-based AI company that has confidentially filed for a U.S. initial public offering, has long acknowledged that its most advanced models possess sophisticated capabilities that could potentially be misused if not properly controlled. The company previously delayed the broad release of its highly advanced Mythos model due to concerns about its ability to assist with cyber-related activities. Earlier this week, however, Anthropic introduced a public version known as Fable, which it said incorporated enhanced cybersecurity protections and safeguards designed to prevent misuse.

The public rollout was short-lived. On Friday, Anthropic announced that it would suspend access to both its Fable 5 and Mythos 5 models after receiving instructions from the U.S. government. According to the company, federal officials informed Anthropic that they believed a method existed for bypassing, or “jailbreaking,” a key security safeguard designed to prevent users from exploiting the model to identify cybersecurity vulnerabilities.

Anthropic maintained that the discovered bypass only exposed relatively minor weaknesses and argued that similar flaws could already be identified by several publicly available AI models. Nevertheless, the company complied with the government’s directive, emphasizing that national security concerns required immediate action.

Under the order, Anthropic was instructed to block access to the two advanced models for all foreign nationals, regardless of whether they were located inside or outside the United States. Faced with the logistical and operational challenges of implementing such restrictions, the company decided to disable access to the models globally while it assessed the situation and worked with authorities.

Amazon declined to confirm whether Jassy directly participated in discussions with government officials regarding Anthropic’s technology. However, a company spokesperson noted that governments frequently consult major cloud service providers on potential security issues because of their extensive relationships with both private-sector and public-sector clients. The spokesperson added that Amazon does not disclose details of such conversations when they occur.

The controversy has intensified the debate surrounding export controls and the regulation of advanced artificial intelligence systems. According to Anthropic, the restrictions were imposed under U.S. export control authorities, a mechanism typically used to limit the transfer of sensitive technologies that could have implications for national security. The U.S. Commerce Department’s Bureau of Industry and Security, which oversees export controls, has not publicly commented on the matter.

White House adviser David Sacks provided further insight into the administration’s decision, stating on social media that officials issued the export control order reluctantly after Anthropic CEO Dario Amodei allegedly declined to either correct the identified security issue or remove the model from public availability. Sacks indicated that the administration’s goal is not to permanently restrict the technology but rather to encourage the company to address the security concerns.

According to Sacks, government officials hope Anthropic will implement additional safeguards, resolve the vulnerabilities identified by regulators, and ultimately return the Fable model to general public release. The comments suggest that the administration views the restrictions as a temporary measure rather than a long-term policy.

The development has sparked confusion and criticism among some experts who generally support tighter controls on advanced AI technologies. Critics argue that the broad scope of the restrictions may create unintended consequences, particularly because they affect citizens of allied nations as well as countries viewed as strategic competitors.

Several analysts noted that the order appears to prevent foreign nationals from accessing the affected models even if they are employed by Anthropic itself or working within the United States. Critics contend that such limitations could hinder legitimate research and development efforts, potentially slowing innovation while providing little additional security benefit.

The timing of the decision is also noteworthy because it comes as relations between Anthropic and certain parts of the Trump administration appeared to be improving after previous disagreements over AI regulation and safety standards. The latest dispute, however, demonstrates how rapidly evolving AI capabilities continue to challenge policymakers who are struggling to balance innovation, competitiveness, and national security.

The incident is likely to fuel broader discussions about how governments should regulate advanced artificial intelligence technologies as they become increasingly capable. With AI models now able to perform sophisticated coding, cybersecurity analysis, and complex reasoning tasks, regulators around the world are facing mounting pressure to establish rules that prevent misuse without stifling technological progress.

For Anthropic, the suspension of Fable 5 and Mythos 5 represents a major setback at a critical time as the company seeks to strengthen its position in the fiercely competitive AI industry. For the broader technology sector, the episode serves as a reminder that advanced artificial intelligence is becoming not only a commercial and technological issue but also a matter of national security with far-reaching global implications.