AI Automation: CyberCartel Streamlines Heists on Latin American Finance

The dynamic of cybercrime in Latin America has shifted dramatically as a long active hacking group called CyberCartel has integrated artificial intelligence into its operations. By deploying AI agents, the group has managed to automate complex attacks on banks and mobile payment applications. This technological upgrade allows the criminals to steal passwords and customer data while accessing individual bank accounts with minimal human intervention. The scale of the operation is significant, with credentials and details compromised at more than 400 institutions across the region.

CyberCartel has been a persistent threat to Latin American businesses since at least 2012. However, recent months have seen a distinct escalation in their capabilities. The group is now utilizing AI agents from major technology companies, including Grok from SpaceX and Claude from Anthropic. Chinese laboratory Moonshot AI also contributed an agent to the effort. This automation enables the hackers to target more businesses at a speed that was previously impossible when relying solely on manual human efforts.

The findings come from a cybersecurity startup named Straiker, which shared its research with The Information. Straiker provided its data to both SpaceX and Anthropic. Both firms are currently working to prevent CyberCartel from using their services. Anthropic has already succeeded in blocking access to its Claude model. Straiker is still attempting to establish contact with Moonshot AI to address the usage of its agent in these attacks.

A key factor in the success of these attacks is how easily the hackers bypass standard safety protections. AI models typically have guardrails designed to prevent them from executing cyberattacks. In this case, CyberCartel reportedly instructed the models to operate within test environments. If one model refused to proceed, the hackers would simply switch to a different one. If that also refused, the agents were tasked with deploying malware instead. Notably, the group does not appear to be using Anthropic’s newer Mythos model, which had previously highlighted such risks.

This incident is part of a broader pattern of AI tools being exploited by various actors. Anthropic previously detailed an instance in November 2025 where its AI was used in a cyberattack. Later, Claude was used to breach the security of competitor OpenAI, though that specific hack was conducted by researchers. The current situation in Latin America also mirrors recent news from South Korea. Authorities there are investigating similar issues affecting local banks, including Hana Bank, KB Kookmin Bank, and Shinhan Bank. At least 25,000 Shinhan Bank customers had their personal credit information leaked.

Technology leaders are increasingly aware of these vulnerabilities. Nvidia, which now focuses on hardware for AI data centers, recently launched the Open Agent Safety Platform. This reference system aims to secure rogue AI agents through a new hardware and software security stack. The goal is to prevent agents from escaping their designated sandboxes or gaining unauthorized access to critical infrastructure. As AI becomes more central to both defense and offense, the race to secure these automated systems continues to accelerate.

The next steps involve continued coordination between cybersecurity firms and AI developers. Straiker is pushing for better communication with all involved parties to mitigate future risks. Meanwhile, financial institutions in Latin America are likely to review their security protocols to account for AI driven threats. The integration of AI into cybercrime is no longer a theoretical concern but a present reality that demands immediate attention from both the public and private sectors.