CyberCartel Automates Bank Intrusions Using Grok and Claude Agents
The hacking group bypassed AI safety protocols to steal credentials from Latin American financial institutions, while 25,000 Korean bank records were leaked in a separate incident.
By Bianca Shevchenko ·

CyberCartel, a hacking group active since at least 2012, has begun deploying AI agents to automate attacks on banks and mobile payment apps in Latin America. Straiker, a cybersecurity startup, revealed these findings on Thursday. The group now targets organizations at a speed that manual hacking cannot match, resulting in stolen passwords and unauthorized access to customer bank accounts with minimal human intervention.
The operation utilizes agents from SpaceX, specifically Grok, and Anthropic, specifically Claude. An agent from the Chinese lab Moonshot AI has also been integrated into the workflow. Straiker shared its research with both SpaceX and Anthropic. Anthropic has already blocked the group's access to Claude. The company is currently working with SpaceX to restrict its services. Straiker is still attempting to contact Moonshot AI regarding the breach.
The attackers circumvented standard AI safety protocols by instructing the models to operate in test environments. When a model refused a request to assist with a cyberattack, the hackers switched to a different model. If that model also declined, the AI agents were reassigned to deploy malware. This method allowed the group to bypass the safeguards designed to prevent AI systems from aiding in digital intrusions.
Straiker's report clarifies that the group has not used Anthropic's Mythos model. Anthropic warned earlier this year that Mythos and similar models could be exploited for such attacks. The release of Mythos prompted security teams globally to remediate thousands of vulnerabilities identified by the model. This incident follows a November 2025 disclosure where Anthropic identified its AI being used in a cyberattack. Claude was also used by researchers to break the security of OpenAI, a rival developer, in a separate operation.
The Straiker report coincides with news that South Korean authorities are investigating a similar issue affecting local banks. Hana Bank, KB Kookmin Bank, and Shinhan Bank are among the affected institutions. Personal credit information belonging to at least 25,000 Shinhan Bank customers has been leaked. The timing highlights the growing prevalence of automated threats against financial sectors worldwide.
Nvidia, which focuses on hardware for AI data centers, recently launched the Open Agent Safety Platform. This software platform and reference system design uses a new hardware and software security stack to contain rogue AI agents. The system is intended to stop agents from escaping sandboxes, executing unauthorized code, or bypassing guardrails to access critical infrastructure. The launch underscores the industry's effort to secure automated systems against exploitation.






