U.S. Officials Warn Bank CEOs of Cyber Risks from Anthropic AI Model

WASHINGTON — U.S. Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell held an urgent meeting with top bank executives this week to warn about potential cybersecurity risks linked to a new artificial intelligence model developed by Anthropic, according to sources familiar with the matter.

The discussions followed the recent unveiling of Anthropic’s advanced AI system, known as Mythos, which the company has not broadly released due to concerns about its potential to expose previously unknown vulnerabilities in digital systems.

Concerns Over Advanced Cyber Capabilities

Anthropic has indicated that the model possesses the capability to identify and exploit weaknesses across major operating systems and web browsers, raising alarms among policymakers and financial institutions.

Sources said the Treasury-hosted meeting in Washington was intended to ensure that leading banks understand the potential risks posed by such AI systems and take proactive measures to strengthen their cybersecurity defenses.

Anthropic has also been in active discussions with U.S. government officials regarding both the defensive and offensive cyber capabilities of its latest model. A source close to the matter said the company proactively briefed senior officials and key industry stakeholders ahead of the model’s limited release.

Limited Access to the Model

Access to the Mythos model is expected to be restricted to a small group of approximately 40 technology companies, including Microsoft and Google, as Anthropic seeks to manage potential risks associated with its deployment.

Banking Leaders Attend Briefing

According to reports, chief executives from major U.S. financial institutions—including Citigroup, Morgan Stanley, Bank of America, Wells Fargo and Goldman Sachs—attended the meeting. JPMorgan Chase CEO Jamie Dimon was unable to participate, one source said.

Officials reportedly issued invitations while many bank leaders were already in Washington for other engagements, enabling rapid coordination on the issue.

Industry Response

Neither the U.S. Treasury nor Anthropic immediately responded to requests for comment. Goldman Sachs, Wells Fargo and the Federal Reserve also declined to comment on the meeting.

Growing Focus on AI-Driven Cyber Risk

The development highlights increasing concern among regulators and financial institutions about the dual-use nature of advanced AI systems, which can enhance both cybersecurity defenses and offensive capabilities.

As AI models grow more powerful, policymakers are intensifying efforts to ensure that critical sectors such as banking are prepared to address emerging threats while balancing innovation and security.