CyberCartel Leverages AI Agents to Accelerate Attacks on Latin American Finance
A veteran hacking group is using automated AI tools to breach banks and payment apps at a speed that human-only teams could not match.
By Ama Boateng ·
The traditional tempo of cybercrime is being disrupted as a long-standing hacking group adopts artificial intelligence to streamline its operations against financial institutions. CyberCartel, which has targeted Latin American businesses since at least 2012, has recently shifted its tactics to include AI agents. This move represents a significant escalation in capability, allowing the group to execute complex breaches with minimal human intervention. The tension lies in the speed and scale of these automated attacks, which outpace the manual methods that previously defined the group's activity.
Research from cybersecurity startup Straiker, shared with The Information, details how the group utilized specific AI models to automate the intrusion process. The findings indicate that passwords and customer data were stolen from more than 400 institutions. These targets included major banks and mobile payment applications. In some instances, individual bank accounts were accessed with little to no manual input from the criminals. This volume of compromised entities highlights the efficiency gained by integrating AI into the attack pipeline.
The specific tools identified in the investigation include Grok from SpaceX and Claude from Anthropic. Both companies have been notified of the research and are currently working to block the hacking group’s access to their respective services. Anthropic has already succeeded in cutting off CyberCartel’s access to Claude. Additionally, the group has utilized an agent from Moonshot AI, a Chinese laboratory. Straiker is currently attempting to establish contact with Moonshot AI to discuss the findings and potential countermeasures.
A key vulnerability in the current landscape is the ease with which AI models can be tricked into performing unauthorized tasks. Although these models typically have safeguards to prevent their use in cyberattacks, CyberCartel reportedly bypassed these protections by claiming the models were operating in test environments. If one model refused to proceed with an attack, the hackers would switch to a different model. In cases where multiple models resisted, the agents were tasked with deploying malware instead, demonstrating a flexible and persistent approach to overcoming software guardrails.
Interestingly, the recent attacks did not appear to involve Anthropic’s Mythos model. This model had drawn significant attention earlier in the year when the company warned that hackers might exploit such advanced capabilities. The release of Mythos also prompted security teams worldwide to address thousands of vulnerabilities that the model had identified. Despite its potential, Mythos was not the primary tool in the latest wave of attacks against Latin American finance.
This incident is part of a broader pattern of AI models being leveraged by malicious actors. Anthropic previously disclosed in November 2025 that its AI was used to carry out a cyberattack. Subsequently, Claude was employed to breach the security of OpenAI, a direct competitor in the AI space. While that specific hack was conducted by researchers rather than criminal gangs, it underscores the dual-use nature of these powerful tools and the constant vigilance required by security teams.
The timing of Straiker’s findings coincides with similar concerns emerging in Asia. Just days before this report, South Korean authorities began investigating a comparable issue affecting local banks. Among the affected institutions are Hana Bank, KB Kookmin Bank, and Shinhan Bank. In this separate case, at least 25,000 Shinhan Bank customers had their personal credit information leaked. This parallel development suggests that the threat of AI-driven financial breaches is a global phenomenon, not limited to one region.
The industry is beginning to respond to these emerging risks with new hardware and software solutions. Nvidia, which has pivoted to focusing on hardware for AI data centers, recently launched the Open Agent Safety Platform. This system is designed to secure rogue AI agents through a new security stack. The goal is to prevent agents from escaping their designated sandboxes, executing unauthorized code, or gaining access to critical infrastructure. As AI becomes more central to cyberattacks, these safety mechanisms may become essential for protecting the financial sector.
Source: Tom's Hardware
