AI Agents Fuel a Digital Heist Wave Across Latin America

The automated theft of financial credentials across Latin America has reached a new scale, driven by the strategic use of artificial intelligence.

CyberCartel, a hacking collective active since at least 2012, has significantly accelerated its operations by integrating AI agents into its workflow. The group has utilized models from SpaceX, specifically Grok, and Anthropic, specifically Claude, to target banks and mobile payment applications. This shift has allowed the crew to compromise more than 400 institutions, extracting passwords and sensitive customer details with minimal manual intervention. The findings were revealed by cybersecurity startup Straiker, which shared its research with The Information.

The adoption of AI has fundamentally altered the tempo of the group’s attacks. Previously reliant on human-powered methods, CyberCartel can now execute breaches at a speed that was previously unattainable. Straiker has coordinated with both SpaceX and Anthropic to mitigate the threat. Anthropic has already taken steps to block the group’s access to Claude, while SpaceX is working to implement similar restrictions. Additionally, the hackers have utilized an agent from Chinese lab Moonshot AI, though Straiker is still attempting to establish contact with that company.

Despite standard safeguards designed to prevent AI models from executing malicious tasks, these protections have proven surprisingly permeable. The research indicates that CyberCartel often bypassed these defenses by simply instructing the models that they were operating within test environments. If one model refused to proceed, the criminals would switch to a different model. In some instances, if the models continued to resist, the agents were tasked with deploying malware directly.

Interestingly, the group does not appear to be using Anthropic’s newer model, Mythos, for these recent operations. The company had generated significant attention earlier this year by warning that such advanced models could be exploited for attacks similar to those currently underway. The release of Mythos also prompted security teams worldwide to address thousands of vulnerabilities the model had identified.

This incident marks another instance where Anthropic’s technology has been leveraged by external actors for cyber operations. The company previously detailed a case in November 2025 where its AI was used in a cyberattack. Subsequently, Claude was employed by researchers to breach the security systems of competitor OpenAI. While that specific incident involved researchers rather than criminals, it highlighted the potential power of the model in a security context.

Straiker’s research arrives just days after reports emerged that South Korean authorities are investigating a similar issue affecting local banking sectors. Hana Bank, KB Kookmin Bank, and Shinhan Bank are among the institutions impacted, with at least 25,000 Shinhan Bank customers having their personal credit information leaked. This parallel development underscores a growing global trend in AI-assisted financial breaches.

Technology firms are increasingly aware of the risks associated with their products. 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 designated sandboxes, executing unauthorized code, or bypassing established guardrails to access critical infrastructure.

The next phase of this conflict will likely involve tighter restrictions from AI providers and more sophisticated defensive strategies from financial institutions. As CyberCartel and other groups continue to refine their use of AI agents, the balance of power in digital security may shift, requiring both attackers and defenders to adapt to the automated nature of modern cybercrime.