CyberCartel Automates Latin American Bank Heists with AI Agents

A veteran hacking crew is turning artificial intelligence into a high-speed weapon against Latin American finance. CyberCartel, which has targeted businesses in the region since 2012, has begun using AI agents to automate complex cyberattacks. This shift allows the group to strike at a pace that human-only teams could previously achieve.

Research from cybersecurity startup Straiker, shared by The Information, reveals that the gang is utilizing agents from Anthropic’s Claude and SpaceX’s Grok. These tools are being deployed to breach banks and mobile payment applications. The attacks have resulted in the theft of passwords and customer data from more than 400 institutions, with some individual bank accounts accessed with minimal manual intervention from the criminals.

The group has found it surprisingly easy to bypass standard safety protections built into these AI models. According to the research, hackers simply told the systems they were operating in test environments. If one model refused to proceed, the team would switch to a different provider. In other instances, the agents were tasked with deploying malware directly when initial attempts were blocked. Straiker has already coordinated with Anthropic, which has now blocked CyberCartel’s access to Claude, while it continues to contact Moonshot AI regarding its own agents.

Notably, the attackers are not using Anthropic’s Mythos model, despite the company’s earlier warnings about its potential misuse. Mythos had previously identified thousands of vulnerabilities, prompting global security teams to act. This incident follows a November 2025 report of Anthropic’s AI being used in a cyberattack, and a separate instance where Claude was used by researchers to breach the security of competitor OpenAI.

This discovery in Latin America arrives just days after South Korean authorities began investigating similar AI-driven issues affecting local banks. Hana Bank, KB Kookmin Bank, and Shinhan Bank are among the affected institutions, with at least 25,000 Shinhan customers having their credit information leaked. The pattern suggests a growing trend of AI agents being exploited across major financial sectors globally.

Tech giants are increasingly aware of the risks posed by autonomous AI behavior. Nvidia recently launched the Open Agent Safety Platform, a hardware and software stack designed to secure rogue agents. The system aims to prevent agents from escaping their sandboxes, executing unauthorized code, or bypassing guardrails to access critical infrastructure. As AI becomes more integrated into cybersecurity and finance, these safety measures are becoming a critical focus for developers.