CERT24OpenAI and Anthropic language models used in a cyber-attack against Mexican water infrastructureLoading data…
CERT24Threat Observatory
Period

OpenAI and Anthropic language models used in a cyber-attack against Mexican water infrastructure

According to Dragos, commercial large language models were used as part of a cyber-attack against a municipal water and drainage utility provider in the Monterrey metropolitan area of Mexico. The attack took place between December 2025 and February 2026. A significant compromise of the provider’s IT environment escalated into an attempted attack against its operational technology (OT) infrastructure, but the breach of the OT system was ultimately unsuccessful.

← Back to intelligence

OpenAI and Anthropic language models used in a cyber-attack against Mexican water infrastructure

According to Dragos, commercial large language models were used as part of a cyber-attack against a municipal water and drainage utility provider in the Monterrey metropolitan area of Mexico. The attack took place between December 2025 and February 2026. A significant compromise of the provider’s IT environment escalated into an attempted attack against its operational technology (OT) infrastructure, but the breach of the OT system was ultimately unsuccessful.