AI agents aimed at 400 Latin American banks

Hackers known as CyberCartel used a network of AI agents built on Anthropic’s Claude and SpaceX’s Grok against more than 400 targets in Latin America, PYMNTS reported on October 21, citing The Information. The targets included banks and payments companies. The research came from the security startup Straiker.

The agents helped steal passwords and consumer data and reached customer bank accounts with little human effort, the report said. Anthropic said it opened an investigation after the findings and quickly blocked the group’s use of Claude. “Offense is way ahead of defense right now, that’s just how it is,” Straiker CEO Ankur Shah said.

The 400 figure counts targets. The reports do not establish how many intrusions succeeded.

What this means for your bank

Assume credential attacks against your staff and customers now run at machine speed. Move high-risk logins to phishing-resistant MFA.
Watch online banking for automated account access patterns, such as many logins from new devices in a short window.

Sources

Hackers employ Grok and Claude to attack LatAm banks (PYMNTS)pymnts.com ↗

This is for information only and is not legal advice. The details come from one security firm’s research as reported in the press.

Keep reading

An AI-agent campaign targeted South Korean banks

Google gives its strongest model to cyber defenders first

Browse the full Regulatory Radar