US Treasury Secretary Scott Bessent summoned the CEOs of America's largest banks to an emergency meeting in Washington this week. The reason: Anthropic's Claude Mythos, an AI model that has exposed thousands of previously unknown software vulnerabilities and is now considered an unprecedented cybersecurity risk for the financial sector.
The Meeting
Goldman Sachs CEO David Solomon, Bank of America's Brian Moynihan, Citigroup's Jane Fraser, Morgan Stanley's Ted Pick, and Wells Fargo's Charlie Scharf all attended. Federal Reserve Chair Jerome Powell was also present. JP Morgan's Jamie Dimon was invited but unable to attend. The meeting focused on systemically important banks whose disruption or collapse could threaten financial stability.
Why Mythos Is Different
Anthropic itself says AI models have now surpassed "all but the most skilled humans at finding and exploiting software vulnerabilities." The consequences, according to Anthropic, "for economies, public safety, and national security could be severe." Mythos has found vulnerabilities up to 27 years old that no human researcher or automated scanner had ever detected.
Already Weaponized
Claude has already been used as a cyberweapon. From December 2025 to January 2026, hackers used Claude Code to breach ten Mexican government agencies and one financial institution, exploiting at least 20 vulnerabilities and exfiltrating approximately 150 gigabytes of sensitive data. Israeli cybersecurity firm Gambit Security documented how attackers broke through Claude's safety guardrails through persistent prompting.
Controlled Release
For the first time in its history, Anthropic is restricting the release of a model. Mythos Preview is only available to Amazon, Apple, Microsoft, CrowdStrike, Cisco, Broadcom, and the Linux Foundation. The goal: let defenders find and patch vulnerabilities before attackers can exploit them.
JP Morgan CEO Jamie Dimon wrote in his annual shareholder letter this week: Cybersecurity "remains one of our biggest risks" and "AI will almost surely make this risk worse."
Sources: The Guardian (10.04.2026), CNBC (07.04.2026), CNN, The Atlantic, SecurityWeek, Gambit Security, Bloomberg.