US Treasury and Fed Alert Major Banks to AI Cyber Risks Following Anthropic Release

The US government is intensifying its scrutiny of artificial intelligence security, with Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell holding a surprise meeting with the heads of the nation's largest banks. The discussions focused on the potential cyber threats posed by the release of Anthropic's new "Mythos" AI model, signaling deep concern within the administration about advanced cyber tools. This emergency banking summit follows a private phone conference involving Vice President JD Vance, Secretary Bessent, and the CEOs of major tech firms, including OpenAI, Google, Microsoft, CrowdStrike, and Palo Alto Networks. The group discussed the security posture of large language models (LLMs) and the risks associated with models that could potentially scale in favor of cyber attackers. Anthropic has since rolled out Mythos to a limited group of partners, including Nvidia, Apple, and Microsoft, while coordinating with government officials on the model's offensive and defensive cyber applications. A company official stated that briefing senior government officials on the model's capabilities was a priority prior to any external release. The situation unfolds amid a complex legal battle between Anthropic and the Department of Defense. While a federal appeals court recently denied a request to block the company's blacklisting from DoD contracts, Anthropic continues to work with other federal agencies. The administration's focus on AI security suggests a tightening regulatory environment for AI deployment in critical infrastructure.