Anthropic Restricts Access to Powerful AI Model Amid Cybersecurity Concerns

Treasury, Federal Reserve Convene Emergency Meeting with Bank CEOs Over Potential Threat to Financial System

Apr. 13, 2026 at 1:58pm

Got story updates? Submit your updates here. ›

A highly detailed, glowing 3D illustration of a complex network of digital hardware and infrastructure, with neon lights highlighting the intricate cybernetic components. The image conveys a sense of the power and vulnerability of essential digital systems in the face of advanced AI-powered threats.As AI capabilities rapidly evolve, the cybersecurity risks to critical infrastructure have become an immediate concern for policymakers and industry leaders.Washington Today

Anthropic, a leading AI lab, has introduced a new AI model called Claude Mythos that is capable of identifying and exploiting software vulnerabilities across major operating systems and web browsers. However, the company has chosen to tightly restrict access to the model under an initiative called Project Glasswing, sharing it only with a limited group of organizations responsible for critical infrastructure. This unusual move has prompted an emergency meeting between the U.S. Treasury Secretary, Federal Reserve Chair, and CEOs of the country's largest banks to discuss the cybersecurity implications of this powerful AI capability.

Why it matters

Anthropic's decision to restrict access to Claude Mythos suggests that the company recognizes the immediate, practical, and potentially destabilizing impact this AI model could have on global systems. The involvement of the Treasury Secretary and Federal Reserve Chair indicates that this capability is being interpreted as a threat to core economic infrastructure, marking a shift from abstract, long-term AI risk discussions to more immediate concerns about what AI is already capable of.

The details

The Claude Mythos model has uncovered thousands of previously unknown 'zero-day' vulnerabilities in internal testing and early deployments, some of which had existed undetected for decades. Rather than releasing the model to the public, Anthropic has chosen to share it with a limited group of organizations, including Amazon, Cisco, Microsoft, and others, with the goal of patching these vulnerabilities before they can be exploited.

  • Over the past two weeks, Anthropic introduced the Claude Mythos AI model.
  • Within days of the model's introduction, the U.S. Treasury Secretary and Federal Reserve Chair convened an urgent, closed-door meeting with the CEOs of the country's largest banks.

The players

Anthropic

A leading AI lab and for-profit company that has developed the powerful Claude Mythos AI model.

Scott Bessent

The U.S. Treasury Secretary who convened the emergency meeting with bank CEOs.

Jerome Powell

The Federal Reserve Chair who convened the emergency meeting with bank CEOs.

Got photos? Submit your photos here. ›

What they’re saying

“The Treasury Secretary and the Chair of the Federal Reserve do not call emergency meetings with every major bank CEO because of a new software release. They do it when they believe something has the potential to destabilize the financial system.”

— Tom Snyder, Author

What’s next

The judge in the case will decide on Tuesday whether or not to allow Walker Reed Quinn out on bail.

The takeaway

This case highlights growing concerns in the community about repeat offenders released on bail, raising questions about bail reform, public safety on SF streets, and if any special laws to govern autonomous vehicles in residential and commercial areas.