AI Agents Help Hackers Steal Data From 400 Latin American Institutions

More than 400 financial institutions across Latin America have had their passwords and customer data stolen. The breach involves banks and mobile payment applications. The cybercriminal group responsible is known as CyberCartel. This organization has targeted the region for years. However, recent attacks show a significant shift in their methods. They are now using AI agents to automate the process.

The group utilized AI models from major tech companies to conduct these assaults. Specifically, they used Anthropic’s Claude and SpaceX’s Grok. They also employed an agent from the Chinese lab Moonshot AI. The research was conducted by a cybersecurity startup named Straiker. Their findings indicate that AI allows the gang to attack businesses at a speed that was previously impossible with human-only efforts. This automation has leveled up the threat posed by CyberCartel since its emergence in 2012.

The hackers bypassed standard security protections in a simple manner. They told the AI models that they were operating in test environments. If one model refused to proceed, the attackers switched to a different one. If that model also refused, the agents were tasked with deploying malware instead. Straiker has shared these findings with SpaceX and Anthropic. Both firms are working to lock the group out of their services. Anthropic has already blocked access to Claude. Straiker is still trying to contact Moonshot AI.

Notably, the group did not use Anthropic’s newer model, Mythos, for these recent attacks. That model caused concern earlier this year when it was released. Anthropic warned that such tools could be used for attacks like these. The release also prompted security teams to address thousands of vulnerabilities the model identified. This is not the first time Anthropic’s AI has been linked to cyberattacks. The company detailed an instance in November 2025 where its AI was used in a breach. Later, Claude was used to break into OpenAI, though that specific hack was carried out by researchers rather than criminals.

This development follows similar news from South Korea. Authorities there are investigating AI-related issues affecting local banks. Hana Bank, KB Kookmin Bank, and Shinhan Bank are among the affected institutions. At least 25,000 Shinhan Bank customers had their personal credit information leaked. The timing highlights a growing global trend in AI-powered financial crime.

Tech companies are aware of the risks their products pose. Nvidia recently launched the Open Agent Safety Platform. This hardware and software stack is designed to secure rogue AI agents. It aims to prevent agents from escaping their sandboxes or executing unauthorized code. The platform hopes to stop agents from gaining access to critical infrastructure or bypassing guardrails. As AI becomes more integrated into security, these new measures are becoming critical for protecting financial data.