Git Informed
Git
Informed
All newsBriefing
Editorial front page
FinalAI-edited source brief

Amazon's CEO Reportedly Triggered the Government Shutdown of Anthropic's Most Powerful AI Models

Andy Jassy raised security concerns about Fable 5 and Mythos 5 to Trump officials—despite Amazon being Anthropic's largest investor and cloud partner.

Published 8 sources0 Reddit7 web82% confidence

What matters

  • Amazon CEO Andy Jassy reportedly told Trump officials that Fable 5 could be used for cyberattacks, triggering a Commerce Department export control ban on Fable 5 and Mythos 5 on June 12, 2026.
  • Amazon is Anthropic's largest investor—$4 billion initially in 2023, plus a $25 billion tranche in April 2026, bringing total committed capital to roughly $33 billion—and its primary cloud infrastructure partner.
  • Anthropic characterized the jailbreak technique as identifying a 'small number of previously known, minor vulnerabilities' that other frontier models can also find, with limited utility workarounds.
  • Cybersecurity expert Katie Moussouris reportedly questioned the severity of the jailbreak findings after reviewing the actual report.
  • The shutdown affects all users globally and also impacted Amazon's own AWS platform, exposing a governance gap with no framework for managing hyperscaler-investor conflicts in AI.

What happened

On June 12, 2026, the US Commerce Department issued an export control directive at approximately 5:21 PM ET that forced Anthropic to shut down its two most powerful AI models—Fable 5 and Mythos 5—for every customer worldwide. The trigger, according to reporting by The Wall Street Journal, The Information, and TechCrunch, was Amazon CEO Andy Jassy himself.

Jassy reportedly told Treasury Secretary Scott Bessent and other senior Trump administration officials that Amazon security researchers had used Fable 5 to obtain information that could be used in cyberattacks. The government then imposed an export control ban on both models. TechCrunch reported on June 13 that Jassy "may have been the source" of the security concerns—hedged language that The CyberSignal and other outlets have preserved, as Amazon has neither confirmed nor denied the specifics on the record.

An Amazon spokesperson said it is "not uncommon for governments to seek our counsel on potential security risks" but declined to share details of the discussions. The spokesperson also pointed to an AWS status update confirming that Amazon's own cloud platform was affected by the model shutdown.

Anthropic has pushed back on the severity of the findings. According to Gizmodo, Anthropic characterized the reported jailbreak technique as identifying a "small number of previously known, minor vulnerabilities" that other frontier models can also find, and that workarounds are limited in utility. Cybersecurity expert Katie Moussouris, who reportedly reviewed the actual jailbreak report, also questioned the severity of the findings.

David Sacks, Trump's former AI czar who now co-chairs the President's Council of Advisors on Science and Technology, offered a different account, saying a "highly credible trusted partner of both Anthropic and the USG came forward" with concerns.

Why it matters

The revelation adds an extraordinary corporate-relationships dimension to an already significant regulatory action. Amazon is Anthropic's largest investor and primary cloud infrastructure partner. Amazon began investing in Anthropic in 2023 with an initial $4 billion commitment, followed by an additional $25 billion tranche in April 2026 for infrastructure needs, bringing total committed capital to roughly $33 billion. In return, Anthropic committed to $100 billion in cloud spending on AWS.

The company that bankrolls Anthropic's infrastructure is also the one that reportedly told the government its models are dangerous enough to shut down globally. In any other industry, the largest investor lobbying the government to restrict its own portfolio company's flagship product would be a scandal. In AI, it's a data point about how the geopolitical and competitive layers actually work.

The episode also exposes a governance gap: the AI industry has no established framework for managing conflicts between hyperscaler-investors and the AI labs they fund, distribute, and simultaneously compete with. Anthropic's own published safety research and capability assessments reportedly gave the government the technical vocabulary to justify the shutdown—transparency turned against its creator.

For Anthropic's anticipated fall 2026 IPO, investors must now assess whether Amazon will behave as an aligned long-term partner or as a party whose interests can diverge in high-stakes situations.

Public reaction

No strong public signal from Reddit or community discussion was available at the time of this report. Analyst commentary has centered on the paradox of an investor sabotaging its own portfolio company, skepticism about the severity of the reported cyberattack threat, and speculation about Amazon's strategic motivations.

What to watch

  • Whether Anthropic formally challenges the export control directive or negotates a path to reinstating Fable 5 and Mythos 5.
  • Which other tech firms were part of the reported coalition of five that allegedly joined Amazon's appeal to the administration.
  • How Anthropic's IPO filings address the Amazon relationship and export control exposure as material risk factors.
  • Whether the export control action extends scrutiny to downstream users who deployed or integrated the restricted models.

Sources

Public reaction

No Reddit or public discussion data was available at the time of this report. Analyst commentary has centered on the paradox of an investor sabotaging its own portfolio company, skepticism about the severity of the reported cyberattack threat, and speculation about Amazon's strategic motivations.

Signals

  • Skepticism about the severity of the reported cyberattack threat
  • Concern about hyperscaler-investor conflicts of interest
  • Speculation about Amazon's strategic motivations

Open questions

  • Was the jailbreak research genuinely dangerous, or was it overstated for competitive or regulatory leverage?
  • Which other tech firms were part of the reported coalition of five?
  • Will Anthropic challenge the export control directive?
  • How will Anthropic's fall 2026 IPO filings address the Amazon relationship as a material risk?

What to do next

Developers

Audit any production dependencies on Fable 5 or Mythos 5 APIs and prepare fallback plans for alternative models.

Both models are offline globally with no confirmed timeline for return, creating immediate availability risk for applications built on them.

Founders

Reassess single-vendor AI dependencies and diversify model providers to reduce exposure to sudden regulatory shutdowns.

The government forced a total global shutdown of two frontier models overnight—any startup relying on one provider's flagship model faces existential risk.

PMs

Review your AI provider's investor and government relationships as part of vendor risk assessment, not just technical capabilities.

Amazon's role in triggering the shutdown shows that a provider's commercial partners can materially affect product availability through regulatory channels.

Investors

Scrutinize Anthropic's fall 2026 IPO filings for risk disclosures related to the Amazon relationship and export control exposure.

The largest investor in Anthropic triggered a government shutdown of its flagship products—this is a material governance risk that IPO investors must price in.

Operators

Document your organization's use of Anthropic models and prepare compliance records in case export control scrutiny extends to downstream users.

The Commerce Department directive was an export control action, which could carry implications for organizations that deployed or integrated the restricted models.

Testing notes

Caveats

  • This is a regulatory and corporate governance story, not a testable product release. Fable 5 and Mythos 5 are currently offline and inaccessible.