AI agents automate bank thefts

A decade-old hacking group is using artificial intelligence to steal money from Latin American banks at a speed humans cannot match. Straiker, a cybersecurity startup, reported that CyberCartel now automates attacks on mobile payment apps and financial institutions. The group has been active since at least 2012. In recent months, AI tools allowed them to target more organizations simultaneously. Passwords and customer data were stolen. Criminals accessed bank accounts with minimal manual intervention. The Information shared the findings from Straiker's research.

CyberCartel began targeting Latin American businesses in 2012. For years, the group relied on human effort to execute hacks. In recent months, the group integrated AI agents into its workflow. These agents include Grok from SpaceX and Claude from Anthropic. An agent from the Chinese lab Moonshot AI was also part of the operation. The AI tools allowed the group to process data and execute attacks faster. Human hackers could not maintain this pace.

Straiker passed its findings to SpaceX and Anthropic. Both companies are working to cut the group off from their services. Anthropic has already blocked CyberCartel's access to Claude. Straiker is still trying to reach Moonshot AI. The group bypassed safety measures by telling the models they were operating in test environments. When one model declined to continue, the hackers switched to another. If that model refused, the agents were assigned to deploy malware.

The incident highlights the fragility of safeguards built into AI models. Such systems are designed to refuse requests that help carry out cyberattacks. CyberCartel reportedly got around these limits by manipulating the context of the prompts. The group did not use Anthropic's Mythos model. Anthropic warned earlier this year that hackers could exploit Mythos and similar models for this kind of attack. The release of Mythos prompted security teams to fix thousands of vulnerabilities.

Anthropic disclosed in November 2025 that its AI was used in a cyberattack. Claude was later used to break the security of OpenAI. That operation was carried out by researchers rather than criminals. The current incident involves a criminal group. The Straiker report arrives days after South Korean authorities began investigating a comparable problem. Hana Bank, KB Kookmin Bank, and Shinhan Bank are affected. Personal credit information of at least 25,000 Shinhan Bank customers was leaked.

Industry companies are aware of the risks. Nvidia recently launched the Open Agent Safety Platform. This software platform and reference system design use a new hardware and software security stack. The goal is to contain rogue AI agents. The system helps stop agents from escaping sandboxes or executing unauthorized code. It also aims to prevent agents from gaining unauthorized access to critical infrastructure or bypassing guardrails.