Hacking group automates attacks on Latin American banks using AI agents
Researchers say a long-standing cybercriminal group is using large language models to steal credentials from over 400 financial institutions with minimal human effort.
By Kateryna Melnyk ·
Financial institutions across Latin America have faced a new wave of automated digital intrusions driven by artificial intelligence. A hacking collective known as CyberCartel has been leveraging AI agents to streamline its attacks on banks and mobile payment platforms, a development that marks a significant shift in how cybercriminals operate in the region.
According to findings from the cybersecurity startup Straiker, the group has successfully compromised more than 400 institutions. The breaches resulted in the theft of passwords and customer data, while also allowing unauthorized access to individual bank accounts. The research, recently shared by The Information, highlights how the integration of AI has reduced the manual labor required for these complex heists.
CyberCartel has been active in the region since at least 2012, but its recent adoption of AI tools has significantly increased its operational speed. The group utilizes agents from major tech firms, including SpaceX’s Grok and Anthropic’s Claude, to execute attacks that would have been too time-consuming for human-led teams. Straiker has notified both companies about the usage, and both are currently working to restrict CyberCartel’s access to their respective services. Anthropic has already managed to block the group’s use of Claude, while Straiker is still attempting to contact Moonshot AI, whose agents were also identified in the research.
The effectiveness of these attacks raises questions about the robustness of standard AI safety protocols. Although AI models typically include safeguards to prevent their use in cyberattacks, CyberCartel reportedly bypassed these protections by instructing the models that they were operating in test environments. When one model refused to proceed, the hackers would switch to a different one. If resistance persisted, the AI agents were tasked with deploying malware directly.
Interestingly, the group does not appear to be using Anthropic’s newer model, Mythos, for these recent campaigns. Anthropic had previously warned that such advanced models could be exploited for similar attacks, a concern that led security teams worldwide to address thousands of vulnerabilities identified by the system earlier this year. This incident follows other recent examples of Anthropic’s technology being used by external parties, including a November 2025 report on AI-driven cyberattacks and a separate breach of OpenAI’s security by researchers.
The discovery in Latin America comes shortly after South Korean authorities began investigating similar AI-assisted intrusions at local banks. Institutions including Hana Bank, KB Kookmin Bank, and Shinhan Bank were affected, with the latter reporting that personal credit information for at least 25,000 customers had been leaked. These parallel developments suggest that the use of AI in financial sector attacks is a growing global trend rather than an isolated regional issue.
Major technology companies are increasingly aware of the risks associated with their products. Nvidia, which has shifted its focus toward AI data center hardware, recently launched the Open Agent Safety Platform. This system is designed to secure rogue AI agents through a new combination of hardware and software security. The platform aims to prevent agents from escaping their designated environments, executing unauthorized code, or bypassing established security guardrails, offering a potential solution to the vulnerabilities exploited by groups like CyberCartel.
Source: Tom's Hardware
