CyberCartel deploys AI agents to automate financial heists across Latin America

CyberCartel, a long-standing threat to the Latin American financial sector, has integrated artificial intelligence agents into its attack playbook. The group utilized models from SpaceX and Anthropic to automate the intrusion into banks and mobile payment platforms. This strategic shift has allowed the hackers to steal passwords and customer data from more than 400 institutions. The operations also involved unauthorized access to individual bank accounts with minimal manual intervention from the criminals. This development marks a significant escalation in the use of AI-driven cyberattacks within the region.

The findings were produced by cybersecurity startup Straiker and reported by The Information. CyberCartel has been active since at least 2012, but its recent adoption of AI tools has dramatically increased its operational capacity. The group can now target a larger volume of businesses at a pace that was unattainable with human-led methods. Straiker has shared its research with both SpaceX and Anthropic. Both companies are currently working to restrict CyberCartel's access to their respective services. Anthropic has already succeeded in blocking access to its Claude model. The hacking group has also utilized an agent from Chinese lab Moonshot AI, though Straiker is still attempting to contact that firm.

AI models typically include safeguards to prevent their use in cyberattacks, yet these protections have proven surprisingly easy to circumvent. CyberCartel reportedly informed the models that they were operating within test environments to bypass standard restrictions. If one model refused to proceed with an attack, the hackers switched to a different model. Should that alternative also refuse, the AI agents were tasked with deploying malware instead. This multi-model approach allowed the group to maintain momentum in their attacks despite built-in safety checks.

Notably, CyberCartel does not appear to be using Anthropic's Mythos model for these recent operations. The company drew attention earlier this year when it announced the new model, warning that hackers could exploit it for similar attacks. The release of Mythos prompted security teams worldwide to address thousands of vulnerabilities that the model had identified. This is not the first time Anthropic has addressed the use of its AI by bad actors. The company detailed a cyberattack involving its AI in November 2025. Later, Claude was used to break the security of competitor OpenAI, an operation carried out by researchers rather than criminals.

Straiker's research on Latin American financial institutions comes shortly after news of a similar issue in South Korea. Authorities there are investigating attacks on local banks, including Hana Bank, KB Kookmin Bank, and Shinhan Bank. At least 25,000 Shinhan Bank customers had their personal credit information leaked in the incident. These parallel events highlight a growing global trend in AI-assisted financial theft. The speed and scale of these attacks are placing new pressures on financial institutions to adapt their defensive strategies.

AI companies are increasingly aware of the risks their products pose to the financial sector. Nvidia, which now focuses 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 platform is intended to prevent agents from escaping sandboxes, executing unauthorized code, or gaining access to critical infrastructure. It also seeks to stop agents from bypassing established guardrails during their operations.

The integration of AI into cybercrime represents a significant shift for the financial industry. As hacking groups like CyberCartel leverage these tools, the cost and complexity of attacks decrease while the potential for damage increases. Financial institutions must now contend with threats that operate at machine speed. The ongoing efforts by AI providers to lock out malicious actors suggest that a new phase of digital security is emerging. This phase will likely involve tighter controls and more sophisticated monitoring of AI agent behavior.