The "Mastermind" Behind the Global Delisting of Anthropic's Mythos and Fable Models: Amazon CEO, a "Major Shareholder"

Anthropic's AI models were halted by the U.S. government within days of their public release, with the trigger coming from its major investor, Amazon.

According to The Wall Street Journal, Amazon CEO Andy Jassy called high-ranking officials including U.S. Treasury Secretary Bessent, stating that its researchers had bypassed the safety guardrails of Anthropic's flagship model, Fable 5, using specific prompts to obtain sensitive information that could be used for cyberattacks. The White House subsequently convened an emergency meeting, and the Department of Commerce ultimately implemented export controls on the Fable and Mythos models, prohibiting access by overseas users. Anthropic subsequently suspended all user access to the two models to comply with regulatory requirements.

This control not only casts a shadow over Anthropic's prospects as it prepares for an IPO, but also reflects the hidden rivalry among tech giants, the U.S. government's anxiety over the safety of frontier AI, and the long-accumulated tension between Anthropic and the Trump administration.

Amazon's Late-Night Call Ignites a Chain Reaction in AI Controls

According to Axios, the incident began late Thursday night local time. Amazon called U.S. government officials and submitted a report, stating that its researchers had successfully breached the security restrictions of Anthropic's Mythos model, accessing content deemed a threat to national security.

Additionally, according to The Wall Street Journal, Amazon's report showed that under the guidance of specific prompts, the Fable 5 model was able to uncover security vulnerabilities in at least four software programs, information typically blocked by the model's safety mechanisms.

Andrew Morris, founder of cybersecurity company GreyNoise Intelligence, pointed out that this information is still quite far from truly dangerous cybersecurity intelligence, and many public tools can also discover similar vulnerabilities. Furthermore, there is no evidence that Amazon researchers breached the deeper layers of protection in the Fable model that prevent vulnerability information from being converted into executable attack code.

Anthropic responded that the vulnerabilities identified by Amazon were relatively basic issues, stating that the company has established adequate security safeguards and always prioritizes safety.

Negotiations Fail Early Friday Morning; Trump Signs Export Controls

According to Axios, government officials held several hours of negotiations with Anthropic early Friday morning local time, requesting that it voluntarily delist its latest models, but no agreement was reached.

According to The Wall Street Journal, Anthropic CEO Dario Amodei further deepened the doubts of some government officials during a call on Friday. Officials believed that Anthropic displayed an unwillingness to cooperate with government security experts to resolve the issues, which corroborated the government's long-standing judgment that the company lacked credibility in safety risk management.

Ultimately, despite reservations about restricting innovation, President Trump signed off on the control action. David Sacks, White House AI Advisor and venture capitalist, stated on social media that this restrictive measure was an "reluctant" move, and that the government's "hope now is for Anthropic to fix the security issues, for the export controls to be lifted, and for the Fable model to be publicly released again."

Notably, Anthropic had notified the U.S. government multiple times before the release of the Fable model and collaborated with key government AI testing agencies, adopting a phased open strategy. It provided preview access primarily to tech companies like Amazon to facilitate software vulnerability screening before wider public release.

Intertwined Interests: Amazon as Both Shareholder and "Whistleblower"

The complexity of this incident lies in the fact that Amazon is both a major investor in Anthropic and its chip supplier, while also deploying Anthropic's most premium models in its own software vulnerability identification business.

According to The Wall Street Journal, citing some insiders, Amazon CEO Jassy's original intent in calling government officials was to issue a general security warning, but the situation rapidly escalated into a comprehensive ban on overseas users by the Department of Commerce. An Amazon spokesperson stated that as a leading cloud service provider serving a large number of public and private clients, it is not uncommon for the government to seek its opinion on potential security risks, and that the company does not disclose details of such discussions.

Some analysts pointed out that the White House's distrust of Anthropic has long existed, involving the company's connections with liberal political donors, its public warnings about the dangers of AI, and its previous hiring of numerous Biden administration officials.

Valuation Under Pressure, User Diversion; OpenAI May Be a Potential Beneficiary

The implementation of these controls at a critical juncture for Anthropic's IPO sprint is particularly sensitive timing.

Reportedly, the company initially planned to go public as early as this autumn. The forced delisting of its flagship model could lead to user migration to other platforms and put pressure on the company's valuation. OpenAI is seen as a potential beneficiary; the company possesses its own powerful cybersecurity models, is gradually opening them to customers, and maintains active communication with the Trump administration.

From a broader perspective, combined with recent executive orders granting security officials greater regulatory authority over AI models, and government discussions regarding potential equity stakes in AI companies, this incident marks a significant escalation in the government's control over the AI industry.