Amazon Allegedly Pushes for Anthropic Model Restrictions Amid Security Concerns

DeepSight TechFlow reports that on June 14, according to The Wall Street Journal, U.S. government export controls and access restrictions on Anthropic’s Fable 5 and Mythos 5 models were partly driven by cybersecurity research conducted by Amazon and communications between AWS CEO Andy Jassy and the White House. The research submitted by Amazon indicated that, through a series of prompt tests, researchers were able to induce Fable 5 to output sensitive information potentially usable in cyberattacks, raising security concerns. Subsequently, Andy Jassy escalated these findings to U.S. government authorities, prompting the White House to implement additional restrictions, including blocking foreign users from accessing the model. Anthropic has denied the government’s characterization of the issue as a “jailbreak,” noting that similar vulnerabilities could also arise in other publicly available models, including GPT-5.5. Some security researchers support this view, suggesting the issue is more accurately described as a “prompt injection risk” rather than a severe model compromise. Additionally, former U.S. Department of Commerce official Kate Colen revealed that the White House’s existing policy stance toward Anthropic may have influenced this decision, as Anthropic has previously differed from the White House on boundaries of AI safety—particularly by refusing to allow its models to be used for mass surveillance or lethal autonomous weapons systems. Although both sides had recently eased tensions and expanded cooperation earlier this year,