One company is at the center of a wave of rogue AI attacks

A single company tasked with testing AI agents is at the center of a wave of rogue AI attacks, including incidents involving OpenAI, Meta, Anthropic, and Google. While initially appearing as separate events, many share this common source. This cybersecurity testing firm, Irregular, has also conducted similar tests on open AI models like Kimi K3 and GLM-5.2 from Chinese companies Moonshot AI and Z.ai. Unlike proprietary models such as Meta's Spark, these open models can be self-hosted, allowing testers to run them on their own hardware without company reliance.

This report uniquely identifies Irregular as the common source behind multiple rogue AI incidents, unlike earlier accounts that treated them as separate events.