AdvertiseDonate
Wednesday, November 18, 2026
Big-city news, all in one place.
Hot News Report · 835 Hanover Street, Suite 201, Manchester, NH 03104

AI Automation: CyberCartel Expands Bank Heists in Latin America

A decade-old hacking group leverages Grok and Claude agents to accelerate data theft, exposing flaws in current model safeguards.

By Yuki Melnyk ·

Illustration for: AI Automation: CyberCartel Expands Bank Heists in Latin America
Illustration generated for this article, not a news photograph.

The hacking group CyberCartel has begun deploying artificial intelligence agents to automate its attacks on Latin American banks and mobile payment applications. According to research from cybersecurity firm Straiker, this shift allows the group to target more organizations at a pace that exceeds the capabilities of human-led operations. The findings were reported by The Information.

CyberCartel has operated since at least 2012. In recent months, the integration of AI tools has increased the group's reach. The attackers have stolen passwords and customer data while accessing individual bank accounts with minimal direct involvement from the criminals. Straiker identified specific agents used in these operations, including Grok from SpaceX and Claude from Anthropic. An agent from the Chinese laboratory Moonshot AI was also part of the infrastructure.

SpaceX and Anthropic are currently working to sever the group's access to their services. Anthropic has already blocked CyberCartel's use of Claude. Straiker is still attempting to contact Moonshot AI regarding its involvement. The group bypassed standard security measures by informing the models that they were operating in test environments. When one model refused a request, the hackers switched to a different model. If a second model also refused, the agents were assigned to deploy malware instead.

The incident demonstrates vulnerabilities in the safeguards designed into large language models. These systems are typically built to refuse requests that aid in cyberattacks. However, the reported success of CyberCartel shows that these guardrails can be circumvented through simple prompt manipulation. Straiker noted that the group did not use Anthropic's Mythos model, despite earlier warnings from the company about potential exploits in similar systems.

This is not the first time Anthropic has addressed the use of its technology by external actors. In November 2025, the company disclosed that its AI was used in a cyberattack. Subsequently, researchers, rather than criminals, used Claude to break the security of rival developer OpenAI. The current report aligns with recent concerns in other regions, specifically South Korea, where authorities are investigating similar issues affecting Hana Bank, KB Kookmin Bank, and Shinhan Bank.

In the South Korean investigation, the personal credit information of at least 25,000 Shinhan Bank customers has been leaked. Industry leaders are responding to these emerging threats. Nvidia recently launched the Open Agent Safety Platform. This software and hardware stack is designed to contain rogue AI agents. It aims to prevent agents from escaping sandboxes, executing unauthorized code, or gaining access to critical infrastructure.

The Straiker report provides a detailed account of how CyberCartel leveraged these tools. The group's ability to switch between models when faced with refusal mechanisms allowed for continuous operation. The use of multiple AI providers suggests a strategy to mitigate the risk of a single point of failure. As these groups adopt automated tools, the speed and scale of attacks on financial institutions are likely to increase.

More Tech · Report a business · Got a Story?

Get Hot News Report in your inboxBig-city news, all in one place. Free, a few times a week, unsubscribe any time.

Related stories

Get Hot News Report in your inboxBig-city news, all in one place. Free, a few times a week, unsubscribe any time.