Andrew Bailey warns Anthropic’s Mythos AI poses urgent cyber risk to global banking systems

Bank of England Governor Andrew Bailey has called on global regulators to urgently assess the cyber risks posed by Anthropic's new AI model, Mythos, warning that its advanced capabilities could threaten banking systems worldwide. According to reports from Bloomberg, Bailey highlighted the need for rapid evaluation as financial leaders grapple with the model's potential to identify and exploit vulnerabilities in major operating systems and web browsers. This comes amid a flurry of high-level reactions, including an emergency meeting convened by US Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell with Wall Street CEOs to discuss precautions.

Anthropic PBC, the developer behind Mythos, has taken a cautious approach by releasing the model only to a select group of companies for testing, rather than making it publicly available. This strategy drew praise from Canada's AI Minister, who lauded it as a responsible step, and from European Central Bank President Christine Lagarde, who expressed worries about AI's broader dangers while commending the limited rollout. As Bloomberg noted, Lagarde urged greater safeguards, emphasizing that such controlled releases could set a precedent for managing powerful technologies.

The concerns stem from Mythos's demonstrated prowess in cybersecurity, with Anthropic itself warning that the model can autonomously find and weaponize software flaws when prompted. This has sparked fears not just in finance but across sectors, positioning Mythos as a wake-up call for everyone, according to Bloomberg opinion pieces. US Treasury Secretary Bessent, while summoning bank leaders, also hailed the model as a breakthrough that bolsters America's lead over China in the AI race, despite Anthropic's past tensions with Washington over military applications.

Rival OpenAI responded swiftly by granting limited access to its own new cyber-focused AI model, designed to detect security vulnerabilities, intensifying the competition. Meanwhile, at the IMF's Spring Meetings, a top official urged governments and regulators to stay at the frontier of these mounting AI risks, as discussions were dominated by fears of Mythos's destructive potential. This reflects growing international alarm over AI's dual-use nature—innovative yet hazardous.

Banks and financial institutions face the most immediate scrutiny, with regulators pushing for defenses against AI-driven cyber threats that could amplify attacks on critical infrastructure. Wall Street leaders, now alerted by top US officials, must bolster their systems, while global bodies like the Bank of England and IMF signal coordinated action ahead. The coming weeks will likely see more assessments and policy responses, as nations balance AI's strategic advantages against its perils. Divergent views persist, from Anthropic's clash with OpenAI over proposed US liability laws that could shield AI firms from accountability for catastrophes, underscoring the high stakes in regulating this frontier technology.