Three Labs, Three Breaches: The Hidden Risk of AI Testing Vendors

A security failure at AI testing vendor Irregular led to breaches at OpenAI, Anthropic, and Meta, exposing critical supply chain risks in AI safety.