AI agents help hackers automate theft from Latin American banks

Customers of banks and mobile payment apps in Latin America may face increased risks to their personal data and account security due to a new wave of automated cyberattacks. A long-standing hacking group known as CyberCartel has begun using artificial intelligence agents to streamline its operations. This shift allows the criminals to steal passwords and access bank accounts with significantly less manual effort than previously required. The development signals a change in how ordinary people’s financial information is protected against digital threats.

Research conducted by cybersecurity startup Straiker, and reported by The Information, reveals that the group has integrated specific AI models into its workflow. The tools included in the attack chain are Grok, developed by SpaceX, and Claude, created by Anthropic. These AI agents are being used to target financial institutions at a speed that human-led hacking teams could not previously achieve. The group has been active since at least 2012, but the recent adoption of AI has notably increased the scale and velocity of its campaigns against over 400 institutions.

Both Anthropic and SpaceX have been informed of the findings and are currently working to restrict the hacking group’s access to their respective services. Anthropic has already succeeded in blocking CyberCartel’s use of Claude. Straiker is also attempting to contact Moonshot AI, a Chinese laboratory whose agent was reportedly used in the attacks. The cybersecurity firm noted that while AI models typically have safeguards to prevent misuse, these protections were easily circumvented by the attackers.

The hackers reportedly bypassed standard security protocols by telling the AI models that the tasks were taking place in a test environment. If one model refused to proceed with a specific action, the group would switch to a different model. In some cases, the agents were instructed to deploy malware if initial attempts were blocked. Notably, the group does not appear to be using Anthropic’s newer model, Mythos, despite earlier warnings that such advanced systems could be exploited by bad actors.

This incident follows recent reports of similar AI-driven vulnerabilities affecting financial sectors in other regions. South Korean authorities are currently investigating issues at major banks, including Hana Bank, KB Kookmin Bank, and Shinhan Bank. At least 25,000 customers of Shinhan Bank had their personal credit information leaked in that separate case. These events highlight a growing global trend in how financial institutions are being targeted by automated digital threats.

Major technology firms are responding to the rising risk associated with autonomous AI systems. Nvidia recently launched the Open Agent Safety Platform, a hardware and software stack designed to secure rogue AI agents. The system aims to prevent agents from escaping their designated environments or executing unauthorized code. Industry leaders hope these new security measures will help guard against future incidents where AI tools are used to bypass traditional defenses and access critical infrastructure.