Anthropic is preparing to brief the Financial Stability Board on cybersecurity vulnerabilities identified by its unreleased AI model Mythos, according to a Financial Times report citing people familiar with the matter.
The discussions are expected to involve finance ministries and central banks represented within the Financial Stability Board, the global regulatory body responsible for coordinating financial rules across G20 economies. The meeting reportedly follows a request from Bank of England Governor Andrew Bailey, who currently chairs the organization.
Anthropic’s Mythos Preview model, announced last month but not yet publicly released, was designed specifically for cybersecurity applications. According to the company, the system can identify decades-old vulnerabilities in web browsers, infrastructure, and enterprise software systems.
The model has drawn significant attention inside cybersecurity and financial policy circles because of concerns that highly capable AI systems could both strengthen cyber defense and accelerate offensive cyberattacks. Financial institutions remain heavily dependent on aging technology infrastructure, making the sector particularly vulnerable to newly discovered attack paths.
Growing Financial Stability Concerns
Bailey publicly warned in April that advanced cybersecurity-focused AI models could create systemic risks if they dramatically improve the ability to identify exploitable weaknesses across critical systems. Speaking at an event at Columbia University, he said Mythos raised questions about whether AI could fundamentally alter the cyber risk landscape by automating vulnerability discovery at unprecedented scale.
The Financial Stability Board’s involvement signals growing concern among regulators that frontier AI systems may evolve into financial stability risks rather than purely technological or operational issues. Regulators increasingly worry that AI-assisted cyberattacks could target banking infrastructure, payment systems, trading platforms, and other critical components of the global financial system.
Anthropic has not publicly released Mythos and currently limits access to a small number of partners involved in cybersecurity testing and infrastructure defense. The restricted rollout reflects broader industry caution surrounding advanced cybersecurity AI systems capable of autonomous vulnerability analysis.
The Expanding AI Governance Debate
The concerns also emerge as governments worldwide debate whether advanced AI models should undergo mandatory safety reviews before deployment. Policymakers in the United States, Europe, and Asia are increasingly evaluating frameworks for governing frontier AI systems with potentially high-impact cyber capabilities.
The briefing highlights how AI governance discussions are expanding beyond issues such as misinformation and labor disruption into national infrastructure security and financial stability. Central banks and financial regulators are paying closer attention to how AI systems could affect systemic resilience, particularly as banks and enterprises integrate AI into core operational workflows.
Anthropic has increasingly positioned itself as a leading developer of enterprise and cybersecurity-focused AI systems. The company recently expanded partnerships across regulated industries including healthcare, finance, and cybersecurity while also launching initiatives such as Project Glasswing, a program designed to help organizations strengthen cyber defense using AI-assisted vulnerability analysis.
The reported FSB discussions also underscore a broader shift in the AI industry as companies developing frontier models become more deeply involved in government policy and regulatory coordination. As cybersecurity-focused AI systems become more advanced, regulators are increasingly treating them as technologies with potential implications for national security, economic stability, and critical infrastructure protection.
