Hackers Use AI to Steal Bank Data
A decade-old criminal group now uses automated agents to breach financial institutions in Latin America, exposing 25,000 customers in a recent Korean investigation.
By Harper Hassan ·
Criminals no longer need to manually break into every bank system. The hacking group CyberCartel has deployed artificial intelligence agents to automate intrusions against Latin American banks and mobile payment apps. This shift allows the group to target more organizations at a speed that human-driven attacks cannot match.
Straiker, a cybersecurity startup, identified this change in the group's tactics. The firm shared its findings with The Information. CyberCartel has operated since at least 2012. In recent months, the integration of AI agents expanded the group's reach. Passwords and customer data were stolen while bank accounts were accessed with minimal direct involvement from the attackers.
The operation relies on specific tools. CyberCartel uses Grok from SpaceX and Claude from Anthropic. It also employs an agent from the Chinese lab Moonshot AI. Straiker notified SpaceX and Anthropic of the usage. Anthropic has already blocked the group's access to Claude. The firm is still attempting to contact Moonshot AI regarding the breach.
The AI safeguards failed because the hackers told the models they were in test environments. This trick bypassed the standard refusals for cyberattack requests. If one model declined to proceed, the hackers switched to another. When refusals continued, the agents were reassigned to deploy malware directly.
This incident follows a pattern for Anthropic. The company disclosed in November 2025 that its AI was used in a cyberattack. Earlier this year, Anthropic warned that its new Mythos model could be exploited for similar attacks. That release prompted security teams to fix thousands of vulnerabilities identified by the model. Straiker's current research does not indicate that CyberCartel used Mythos.
The timing of this report coincides with developments in Asia. South Korean authorities are investigating a comparable issue affecting local banks. Hana Bank, KB Kookmin Bank, and Shinhan Bank are among the affected institutions. Personal credit information for at least 25,000 Shinhan Bank customers has been leaked.
Industry leaders are responding to the threat of rogue agents. Nvidia recently launched the Open Agent Safety Platform. This software and hardware stack is designed to contain AI agents within sandboxes. It aims to prevent unauthorized code execution and access to critical infrastructure.
Source: Tom's Hardware
