Experts Warn of Long-Term Consequences for U.S. Cyber Defenses
The Trump administration's recent crackdown on Anthropic's AI models, Fable 5 and Mythos 5, has sparked widespread concern among cybersecurity leaders. Nearly 150 security experts, including former Facebook security chief Alex Stamos, have signed an open letter urging the administration to reverse its decision, arguing that it sets a dangerous precedent for American AI companies. The dispute began after Amazon CEO Andy Jassy raised concerns about Chinese access to Anthropic's tools, prompting the administration to restrict their usage.
"They've set a precedent that American models can't do defensive security research," Stamos said.
Amazon's concerns reportedly centered on a "jailbreak" vulnerability that allows Fable 5 to write "proofs of concept," a capability commonly used by security teams to identify and fix weaknesses. However, cybersecurity experts, including Katie Moussouris of Luta Security, argue that the issue does not involve mass exploitation but rather prompts designed to aid defensive work. Moussouris warned that stripping such capabilities would render AI models less useful for cyber defenders.
The administration's response has raised fears that U.S. adversaries, particularly China, could gain a strategic advantage. Stamos cautioned that Chinese AI developers and government-backed hacking groups are unlikely to abandon similar tools, leaving American defenders at a disadvantage. Researchers are now questioning whether the government is creating unwritten rules that could stifle AI security research in the U.S.
Mixed Signals from the Trump Administration
The fallout comes as the Trump administration attempts to establish a vulnerability clearinghouse under its AI security executive order. However, critics note that recent White House departures and the sidelining of top cyber agencies have left the administration with limited cybersecurity expertise. Stamos described the situation as a "huge threat to American dynamism" and a departure from the nation's traditional strengths in innovation and defense.
As the debate continues, the administration's handling of the Anthropic case could have lasting implications for the future of AI cybersecurity in the United States.