The AI industry experienced a massive shockwave when Anthropic abruptly shut down worldwide access to its two most advanced models, Claude Fable and Mythos 5. Everything pointed to last-minute findings by an external team that led to the decision. Now, a fresh report suggests the catalyst for the Claude Fable 5 and Mythos 5 ban came from Amazon, which ironically is one of Anthropic’s primary financial backers.
According to a Wall Street Journal report, Amazon CEO Andy Jassy personally spoke with senior Trump administration officials, including Treasury Secretary Scott Bessent, to raise urgent cybersecurity warnings.
The paper that started the fire
The friction traces back to an internal research paper compiled by Amazon’s cybersecurity team. The report claims that through specific prompt sequences, researchers could bypass safeguards on Anthropic’s Claude Fable 5 model. This could force the AI to yield information that could assist hackers in executing cyberattacks.
Amazon did not explicitly confirm the private details of these executive meetings. A spokesperson noted that governments frequently seek the cloud provider’s counsel on emerging digital threats. However, the consequences of sharing that data were swift and severe. The White House quickly issued an emergency export control directive aimed directly at Anthropic.
An aggressive government response
The federal order legally bars foreign nationals—both inside and outside the United States—from accessing Fable 5 and Mythos 5. Because Anthropic employs a heavily international research team, the sweeping restriction effectively locked scientists out of working on their own creations. Faced with these strict limitations, Anthropic chose to disable the models globally.
White House adviser David Sacks claimed the administration pushed Anthropic CEO Dario Amodei to deploy an immediate fix or voluntarily pull the models down. According to Sacks, Amodei’s initial refusal prompted the government to issue the export control restrictions reluctantly.
Anthropic pushes back
Anthropic is actively disputing the administration’s dramatic “jailbreak” narrative. In a recent blog post, the startup argued that any minor security vulnerabilities found within Fable 5 are already common knowledge and can be easily uncovered using rival public platforms, such as OpenAI’s GPT 5.5. Independent security experts, including LutaSecurity CEO Katie Moussouris, have supported this stance. Moussouris noted the flagged research does not constitute a true system jailbreak.
The crackdown also highlights lingering political tensions. Anthropic previously clashed with federal agencies over its firm refusal to let its models power mass surveillance programs or lethal autonomous weapons systems.
Policy analysts are already criticizing the messy execution of the ban. Many believe the administration’s hasty move could harm domestic tech development more than it protects it. For now, the future of Anthropic’s flagship models remains completely up in the air.
The post Amazon’s Cybersecurity Team Triggered the Anthropic Claude Fable Ban, Report Claims appeared first on Android Headlines.