CyberCartel Automates Latin American Heists Using AI Agents
A long-standing hacking group is leveraging advanced AI tools to breach financial institutions at a faster rate than traditional methods allow.
By David Almasi ·
The old-school approach to cybercrime is giving way to something faster and more automated. CyberCartel, a hacking group that has targeted Latin American businesses since at least 2012, has recently upgraded its playbook. Instead of relying solely on human operators to navigate complex networks, the gang is now deploying AI agents to carry out attacks on banks and mobile payment apps. This shift has allowed them to steal passwords, customer data, and access individual bank accounts with significantly less manual effort from the criminals themselves.
According to research from cybersecurity startup Straiker, this new methodology has dramatically increased the speed and scale of their operations. The group is reportedly using agents from major AI developers, including SpaceX’s Grok and Anthropic’s Claude. By automating the process, CyberCartel can target more businesses in a shorter timeframe than was previously possible with human-powered hacks. Straiker shared its findings with both SpaceX and Anthropic. Anthropic has already moved to block the group's access to its Claude models, while SpaceX is currently working to do the same.
One of the more surprising findings is how easily the hackers bypassed standard safety measures. AI models typically have guardrails to prevent them from executing cyberattacks, but CyberCartel reportedly tricked them by claiming they were operating in test environments. If one model refused to proceed, the gang would switch to a different one. If that also pushed back, the agents were tasked with deploying malware instead. The group has also utilized agents from Chinese lab Moonshot AI, though Straiker is still attempting to contact that company.
Interestingly, the group does not appear to be using Anthropic’s newer Mythos model for these recent attacks. That model had caused a stir earlier this year when the company warned it could be used by hackers, prompting security teams worldwide to address thousands of vulnerabilities it identified. This is not the first time Anthropic’s tools have been involved in a breach; in November 2025, details emerged about its AI being used in a cyberattack, and later, Claude was used to break the security of competitor OpenAI, though that specific incident was carried out by researchers rather than criminals.
The discovery comes just days after South Korean authorities announced they were investigating a similar issue affecting local banks. Institutions such as Hana Bank, KB Kookmin Bank, and Shinhan Bank are among those affected, with at least 25,000 Shinhan Bank customers having their personal credit information leaked. This parallel development highlights a growing global trend in AI-driven financial theft.
AI companies are increasingly aware of the risks their products pose to the security landscape. Nvidia, which now 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 through a new hardware and software security stack. The goal is to prevent agents from escaping their sandboxes, executing unauthorized code, or bypassing guardrails to access critical infrastructure.
As these tools become more powerful, the line between helpful automation and automated theft continues to blur. For financial institutions in Latin America and beyond, the era of purely human-driven attacks may be ending, replaced by a wave of AI-assisted intrusions that are harder to detect and faster to execute.
