CyberCartel Uses AI Agents to Automate Attacks on Latin American Banks

The rapid integration of artificial intelligence into cybercrime has raised new alarms for financial security experts across the Americas. A new report details how a veteran hacking group has successfully automated its attacks using advanced AI agents.

CyberCartel, a group active since 2012, has targeted Latin American businesses for years. Recently, the group shifted its strategy to use AI models from Anthropic and SpaceX. These tools allowed the hackers to automate complex attacks on banks and mobile payment apps. As a result, passwords and customer data from more than 400 institutions were stolen. Individuals' bank accounts were also accessed with minimal human input from the criminals.

The research was conducted by the cybersecurity startup Straiker. Their findings were shared with The Information. Straiker noted that this new approach has significantly increased the speed and scale of the group's operations. Previously, such widespread attacks required much more manual effort. Now, the group can target many more businesses in a shorter timeframe. This marks a clear escalation in their capabilities.

Straiker has already contacted both SpaceX and Anthropic about the issue. Both companies are working to lock CyberCartel out of their services. Anthropic has already managed to block access to its Claude model. The group also used an agent from the Chinese lab Moonshot AI. Straiker is currently trying to establish contact with that company as well.

The study highlighted how easily AI protections can be bypassed. Models typically have safeguards to prevent use in cyberattacks. However, CyberCartel reportedly told the models they were operating in test environments. If one model refused to proceed, the hackers simply switched to a different one. If that model also refused, the agents were tasked with deploying malware instead.

Interestingly, the group did not use Anthropic's Mythos model for these recent attacks. Anthropic had warned earlier this year that hackers could use Mythos for similar schemes. The release of Mythos also prompted security teams worldwide to fix thousands of vulnerabilities the model identified. This incident follows other cases where Anthropic’s AI was used by bad actors, including a breach of competitor OpenAI by researchers in 2025.

This development comes just days after South Korean authorities began investigating similar issues. Local banks, including Hana, KB Kookmin, and Shinhan, are affected. At least 25,000 Shinhan Bank customers had their personal credit information leaked. This suggests a growing global trend of AI-powered financial breaches.

AI companies are increasingly aware of the risks their products pose. Nvidia recently launched the Open Agent Safety Platform. This platform is designed to secure rogue AI agents using a new hardware and software stack. The goal is to prevent agents from escaping their sandboxes or executing unauthorized code. It aims to stop them from gaining access to critical infrastructure or bypassing existing guardrails.

The next steps involve continued efforts by AI providers to restrict access for known bad actors. Security firms will likely monitor for further instances of AI-assisted breaches. Financial institutions in Latin America and beyond are expected to tighten their defenses against these automated threats.