CyberCartel Deploys AI Agents to Target Latin American Banks

The hacking group CyberCartel has harnessed artificial intelligence agents, including Anthropic's Claude and SpaceX's Grok, to automate attacks against financial institutions, stealing passwords and customer data from more than 400 organizations in the process.

The research, conducted by cybersecurity startup Straiker and shared through The Information, details how the group has targeted banks and mobile payment applications. While CyberCartel has been an active threat against Latin American businesses since at least 2012, the integration of AI tools has recently increased the scale and speed of its operations, allowing the group to compromise accounts with minimal human intervention.

Straiker's findings indicate the group can now attack businesses at a pace that was previously impossible with human-powered hacking. Following the disclosure, both SpaceX and Anthropic have taken steps to restrict the group's access to their platforms. Anthropic has already blocked Claude, while Straiker is attempting to contact Moonshot AI regarding the use of its agent as well.

Despite built-in safeguards designed to prevent AI models from executing cyberattacks, CyberCartel reportedly bypassed these restrictions by claiming to be operating in a test environment. The group also employed a fallback strategy, switching to a different model if the first refused to comply, or instructing the AI to deploy malware if the second model declined as well.

CyberCartel's operations do not appear to involve Anthropic's Mythos model. Anthropic earlier this year warned that models like Mythos could be used for attacks similar to those conducted by CyberCartel, and its release prompted security teams worldwide to address thousands of vulnerabilities it identified.

This is not the first time Anthropic has reported misuse of its technology. The company detailed an incident in November 2025 involving its AI in a cyberattack, and later, Claude was used to break the security of rival firm OpenAI, though that breach was carried out by researchers rather than criminals.

The report on Latin American institutions follows closely on 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, with at least 25,000 Shinhan Bank customers having their personal credit information leaked.

AI companies are taking steps to mitigate these risks. Nvidia recently launched the Open Agent Safety Platform, developing a hardware and software security stack designed to secure rogue AI agents. The system aims to prevent agents from escaping sandboxes, executing unauthorized code, accessing critical infrastructure, or bypassing guardrails.

As AI-powered cyberattacks become a shared concern across financial sectors globally, the focus remains on securing the tools that are increasingly being weaponized against the institutions they were built to assist.