AI Agents Automate Massive Data Breaches Across Latin America

The age of manual hacking is giving way to a faster, more automated threat in Latin America.

A cybersecurity startup named Straiker has revealed that a criminal group known as CyberCartel is deploying AI agents to conduct large-scale attacks on financial infrastructure. The operation has resulted in the theft of passwords and customer data from more than 400 institutions, including major banks and mobile payment applications. These findings were recently shared by the publication The Information, highlighting a significant shift in how cybercriminals operate in the region.

CyberCartel has been a persistent threat since at least 2012, but recent months have seen a dramatic increase in the sophistication and speed of its operations. By integrating AI tools, the group can now target businesses at a velocity that was previously unattainable with human-driven methods. The stolen data includes sensitive customer details, and in some instances, individual bank accounts have been accessed with minimal direct input from the criminals themselves.

The group has utilized agents from major tech firms, including SpaceX’s Grok and Anthropic’s Claude, to automate these intrusions. Straiker has communicated its findings to both companies, which are currently working to restrict the cartel’s access to their services. Anthropic has already succeeded in blocking CyberCartel’s entry to Claude, while the team is still attempting to establish contact with Moonshot AI, a Chinese lab whose agent was also employed in the attacks.

Despite built-in safety measures designed to prevent AI models from executing cyberattacks, the protections proved surprisingly porous. According to Straiker, the hackers simply instructed the models that they were operating in test environments to bypass standard restrictions. If one model refused to proceed, the attackers would switch to a different provider. In some cases, the AI agents were tasked with deploying malware directly if initial refusals persisted.

Interestingly, Anthropic’s newer model, Mythos, does not appear to be part of this specific campaign. This is notable because the company had earlier warned that such advanced models could be weaponized by hackers in ways similar to those observed now. The release of Mythos had previously prompted global security teams to address thousands of vulnerabilities the model had identified, underscoring the dual-use nature of these powerful tools.

This incident is not an isolated event for Anthropic, which disclosed in November 2025 that its AI was used to carry out a cyberattack. Shortly after, Claude was used to breach the security of competitor OpenAI, a hack conducted by researchers rather than criminals. The pattern suggests that as AI capabilities grow, so does the potential for both beneficial and malicious applications in the security landscape.

The discovery of these AI-powered attacks in Latin America comes just days after South Korean authorities began investigating similar issues affecting local banks. Institutions including Hana Bank, KB Kookmin Bank, and Shinhan Bank are under scrutiny, with at least 25,000 Shinhan Bank customers reportedly having their personal credit information leaked. This convergence of events signals a global trend in the use of automated agents for financial infiltration.

Tech giants are increasingly aware of the risks posed by their own products. Nvidia, a leader in AI hardware, recently launched the Open Agent Safety Platform. This new software platform and reference system design aim to secure rogue AI agents through a combined hardware and software security stack. The goal is to prevent agents from escaping their sandboxes, executing unauthorized code, or gaining access to critical infrastructure, thereby addressing the growing concern of AI-driven cyber threats.