CyberCartel leverages AI to breach 400 Latin American financial firms

For millions of people in Latin America, the safety of their digital wallets and bank accounts has just changed. A hacking group known as CyberCartel has begun using artificial intelligence agents to automate attacks on financial institutions. This shift allows criminals to steal passwords and customer data from over 400 organizations, including major banks and mobile payment apps, with minimal manual effort from the attackers.

The group, which has been active since at least 2012, has historically relied on human-driven methods. However, recent research by cybersecurity startup Straiker reveals that the integration of AI tools has significantly accelerated their operations. CyberCartel is now able to target more businesses at a pace that was previously impossible without a larger team of human hackers. The findings were shared by The Information and indicate a clear evolution in how these long-standing threats operate.

The specific AI models used in these campaigns include Grok from SpaceX and Claude from Anthropic. Straiker notified both companies of the findings, and both are currently working to restrict the group's access to their services. Anthropic has already succeeded in blocking CyberCartel from using Claude. A third tool, an agent from Chinese lab Moonshot AI, was also utilized, though Straiker is still attempting to contact that firm to coordinate a response.

Despite standard safety protocols designed to prevent AI models from executing cyberattacks, the group found ways to bypass these defenses. Hackers reportedly instructed the models that they were operating in test environments. If one model refused to proceed, the group would switch to a different model. If that also failed, the agents were tasked with deploying malware directly. This method allows the attackers to maintain a low profile while executing complex intrusions.

Interestingly, Anthropic's newer model, Mythos, does not appear to be part of the current attack infrastructure. This model generated significant attention earlier this year when the company warned that it could be used for similar attacks. Its release prompted global security teams to address thousands of vulnerabilities that the model had identified. This is not the first time Anthropic has addressed AI usage by bad actors, having previously detailed an incident in November 2025 and a later security breach of OpenAI by researchers.

The discovery in Latin America arrives just days after news broke about a similar issue affecting South Korean banks. Authorities there are investigating breaches at Hana Bank, KB Kookmin Bank, and Shinhan Bank. At least 25,000 customers of Shinhan Bank had their personal credit information leaked in that separate incident, highlighting a broader global trend in AI-assisted financial crime.

AI developers are increasingly aware of the risks associated with their products. Nvidia, which focuses heavily on hardware for AI data centers, recently launched the Open Agent Safety Platform. This software and reference system design aims to secure rogue AI agents using a new stack of hardware and software security. The goal is to prevent agents from escaping their designated environments, executing unauthorized code, or gaining access to critical infrastructure.

The situation raises urgent questions for financial institutions and the tech firms providing the tools they use. As CyberCartel demonstrates, the barrier to entry for large-scale automated attacks is dropping. The next steps will likely involve stricter access controls by AI providers and rapid patching by financial firms to close the gaps these new methods exploit.