Bessent Shapes AI Order Over Risks to $283 Trillion Fed Payment Rails
Updated
Updated · Semafor · Jun 26
Bessent Shapes AI Order Over Risks to $283 Trillion Fed Payment Rails
3 articles · Updated · Semafor · Jun 26
Summary
Scott Bessent took a larger role in AI policy this spring after banks warned advanced models could expose Fed payment systems to outages that would disrupt money transfers across the financial system.
An April meeting with Jerome Powell and CEOs of systemically important banks focused on risks tied to Anthropic’s Mythos, pushing Bessent to steer the Trump administration away from a hands-off approach.
The resulting executive order created a voluntary review process for advanced AI models and put Treasury in charge of a new cyber clearinghouse, though implementation could be delayed by talks over bringing Mythos back online.
Fedwire handled more than 54 million transfers worth over $283 trillion in the first quarter of 2026, underscoring why banks and regulators have spent years developing backup transfer options despite limited private-sector alternatives.
Can a 'voluntary' government review truly safeguard trillions in daily transactions from advanced AI threats?
Is the U.S. inadvertently ceding its AI leadership to global competitors by restricting its own top models?
The 2026 AI Financial Crisis: Anthropic’s Claude Mythos, Systemic Risk, and the New Era of Cybersecurity Regulation
Overview
Between April and June 2026, the sudden emergence of Anthropic’s advanced AI model, Claude Mythos, triggered an urgent response from U.S. regulators and financial institutions. Fears of AI-powered cyberattacks on critical financial infrastructure led to immediate action, with the Trump administration shifting its policy and prompting unprecedented collaboration among government, banks, and tech companies. High-level meetings underscored the gravity of the threat, as leaders recognized that powerful AI could both defend and endanger global financial stability. This crisis set off a chain of policy changes and industry partnerships aimed at containing new systemic risks posed by frontier AI.