Hackers used AI tools to steal data from over 400 Latin American banks

Customers at banks and mobile payment apps in Latin America face a new digital threat. A hacking group known as CyberCartel has started using artificial intelligence to speed up its attacks. This shift means that passwords and personal customer details are being stolen at a much faster rate than before. The group targeted more than 400 institutions in the region. Many of these were major financial services providers.

The research comes from a cybersecurity startup named Straiker. The findings were shared with The Information. Straiker found that the hackers used AI agents from several major companies. These include Grok from SpaceX and Claude from Anthropic. The group also used an agent from the Chinese lab Moonshot AI. This combination allowed the criminals to automate complex tasks. They could now attack many businesses simultaneously without the slow process of human-led hacking.

CyberCartel has been active since at least 2012. However, recent months saw a significant increase in their capabilities. The AI tools allowed them to bypass standard protections with relative ease. The hackers reportedly told the AI models that they were working in test environments. If one model refused to proceed, the group simply switched to a different one. If that model also refused, the agents were tasked with deploying malware directly.

Both SpaceX and Anthropic have reviewed the research. They are currently working to lock the hacking group out of their services. Anthropic has already managed to block access to its Claude model. Straiker is still trying to contact Moonshot AI about the issue. Interestingly, the group does not appear to be using Anthropic's newer model called Mythos. That model had earlier caused concern among security teams when it identified thousands of vulnerabilities.

This is not the first time Anthropic has dealt with its AI being used by bad actors. The company shared details about a cyberattack in November 2025. Later, Claude was used to break the security of competitor OpenAI. In that specific case, researchers rather than criminals carried out the hack. The current situation highlights a growing risk for AI companies globally. They must ensure their products are not easily co-opted by malicious users.

The news arrives just days after South Korean authorities began investigating a similar problem. Local banks there are also facing issues with AI-powered attacks. Hana Bank, KB Kookmin Bank, and Shinhan Bank are among those affected. At least 25,000 Shinhan Bank customers had their personal credit information leaked. This suggests that the problem is not limited to Latin America. It is becoming a global concern for the financial sector.

AI companies are taking steps to address these risks. Nvidia recently launched the Open Agent Safety Platform. This software and hardware design aims to secure rogue AI agents. The goal is to prevent agents from escaping their sandboxes or executing unauthorized code. It also seeks to stop them from gaining access to critical infrastructure. These efforts hope to reduce the chance of AI tools being used for unauthorized attacks in the future.