A hacking group known as CyberCartel has started using AI agents to automate cyberattacks on financial services in Latin America. The AI tools include Anthropic’s Claude, SpaceX’s Grok and an agent from Chinese lab Moonshot AI. Straiker, a cybersecurity startup, says the group has taken data from more than 400 institutions, including banks and mobile‑payment apps, and has accessed individual accounts with very little manual effort.
How the AI tools are being used
According to Straiker’s research, the AI agents can generate phishing messages, guess passwords and navigate compromised systems faster than a human hacker could. The group can launch attacks at a scale that was not possible with purely manual methods. Straiker shared its findings with the AI providers, and Anthropic has already blocked the offending accounts from using Claude. SpaceX is also working on a fix for Grok, while Moonshot AI has not yet responded.
Why it matters
For everyday users of banks and payment apps in Latin America, the use of AI means their data can be compromised more quickly and at larger scale. It shows that even widely available AI services can be misused, prompting providers to tighten access controls. Until the fixes are in place, users should remain vigilant for suspicious account activity.
