OpenAI and Anthropic LLMs Used in Critical Infrastructure Cyber-Attack, Warns Dragos

Dragos, an industrial cybersecurity firm, has reported that commercial large language models (LLMs) including OpenAI's GPT models and Anthropic's Claude were used in a cyber-attack on a municipal water utility in Mexico. The firm did not provide specifics on the attack method or attribution, but the incident highlights the growing threat of AI misuse in critical infrastructure.

The use of frontier AI models in an operational technology environment marks a significant escalation. Dragos specializes in securing industrial control systems, and its detection of LLM involvement suggests a new level of sophistication. Water utilities are particularly susceptible due to legacy systems and limited budgets.

This incident is likely to fuel debates about AI regulation and the responsibility of model developers to prevent misuse. While no official response from OpenAI or Anthropic has been noted, the case underscores the dual-use nature of advanced AI. Dragos's warning adds to a growing body of evidence that AI tools are being weaponized by malicious actors.