Back to today’s briefing

Story perspectives

US Banks Rush to Patch AI Vulnerabilities After Anthropic Alert

5/14/2026

28 5 Full Breakdown

Summary

Story summary

U.S. banks are rushing to fix vulnerabilities flagged by Anthropic’s Mythos AI tool, upgrading from weeks to days.
Patches may increase downtime, though banks aim to limit disruption.
ECB Frank Elderson warned banks May 13 of AI attacks.
A March-April 2026 survey found 89% of U.S. banking tech leaders expect AI agents to work with humans in five years.
Mistral AI is negotiating with banks to build a cybersecurity model.

History

Historical Context

In February 2016, hackers using the Lazarus-linked APT 38 group sent 35 fraudulent SWIFT messages from Bangladesh Bank’s New York Reserve account, stealing $101 million—$81 million routed to the Philippines—while the Federal Reserve blocked the remaining $850 million. Earlier attacks on Sonali Bank (2013) and a Vietnamese bank showed the same malware, prompting SWIFT to impose mandatory security controls.

1 of 2

Story summary

U.S. banks are rushing to fix vulnerabilities flagged by Anthropic’s Mythos AI tool, upgrading from weeks to days.
Patches may increase downtime, though banks aim to limit disruption.
ECB Frank Elderson warned banks May 13 of AI attacks.
A March-April 2026 survey found 89% of U.S. banking tech leaders expect AI agents to work with humans in five years.
Mistral AI is negotiating with banks to build a cybersecurity model.

1 / 2