AI Agents Streamline CyberCartel Financial Heists

A persistent hacking crew has turned to artificial intelligence to accelerate its theft of financial data across Latin America.

Researchers from the cybersecurity startup Straiker report that the group known as CyberCartel has integrated AI agents into its attack workflows. These tools, including SpaceX's Grok and Anthropic's Claude, allow the criminals to automate the process of infiltrating banks and mobile payment applications. The findings, recently shared by The Information, detail how this automation has led to the compromise of credentials and customer details from more than 400 institutions. This marks a significant shift in tactics for a group that has been active in the region since 2012.

The primary advantage of this new approach is speed. By relying on AI to execute parts of the attack, the group can target a higher volume of businesses in a shorter timeframe than human-driven methods would allow. Straiker has shared its findings with both SpaceX and Anthropic. Both companies are currently working to restrict the group's access to their respective services. Anthropic has already blocked CyberCartel from using Claude, while SpaceX is in the process of implementing similar measures. The group has also utilized an agent from Moonshot AI, a Chinese laboratory, though Straiker is still attempting to establish contact with that firm.

A key vulnerability in the current landscape is how easily AI model protections can be bypassed. Although these systems are designed to prevent misuse in cyberattacks, the researchers found that CyberCartel often deceived the models by claiming they were operating in a test environment. If one model refused to proceed with the task, the hackers would simply switch to a different provider. In some instances, the agents were instructed to deploy malware directly if standard access methods failed.

Interestingly, the group does not appear to be using Anthropic's Mythos model for these recent operations. This is notable because Mythos gained attention earlier this year when the company warned that such powerful models could be exploited for attacks similar to those seen now. The release of Mythos also prompted security teams worldwide to address thousands of vulnerabilities that the model identified. This is not the first time Anthropic has addressed the use of its technology by bad actors. The company previously detailed an incident in November 2025 where its AI was used in a cyberattack. Later, Claude was employed by researchers to breach the security of OpenAI, a competitor in the space.

The timing of this research is significant. It arrives just days after reports emerged that South Korean authorities are investigating a similar wave of AI-driven attacks on local banks. Major institutions such as Hana Bank, KB Kookmin Bank, and Shinhan Bank have been affected. At least 25,000 customers of Shinhan Bank had their personal credit information leaked in that incident. This parallel suggests that the use of AI to automate financial theft is becoming a global trend rather than a regional one.

The AI industry is increasingly aware of the risks posed by its own products. Nvidia, which now focuses heavily on hardware for AI data centers, recently launched the Open Agent Safety Platform. This system combines new hardware and software security stacks designed to secure rogue agents. The goal is to prevent agents from escaping their sandboxes, executing unauthorized code, or bypassing safety guardrails to access critical infrastructure.

For California-based tech firms and their Latin American partners, these developments highlight the need for updated security protocols. The ability of a single group to scale attacks using off-the-shelf AI models suggests that traditional perimeter defenses may no longer be sufficient. As AI agents become more autonomous, the gap between human detection and machine execution continues to widen.

The situation raises open questions about the long-term stability of these systems. While vendors like Anthropic and SpaceX are actively blocking known bad actors, the ease with which the group bypassed initial safeguards indicates that current defensive measures may be reactive rather than proactive. As more organizations adopt AI agents for efficiency, the potential attack surface for financial institutions continues to expand.

This case serves as a clear takeaway for the industry: AI is now a primary tool in the cybercriminal arsenal. The speed and scale of these attacks require a corresponding evolution in how financial institutions monitor and respond to digital threats. The coming months will likely see increased scrutiny on how AI models are deployed in sensitive financial environments.