Google's threat-intel team disclosed the first AI-developed zero-day in the wild

An unknown actor used an AI system to discover the vulnerability and generate the exploit. State-linked groups in China, Russia, and North Korea are now integrating AI directly into attack workflows. Almost simultaneously, an AI-generated Linux kernel patch — for AMD's Promontory 21 chipset, written by GPT-5.5 via Codex — was submitted upstream, and a malicious Hugging Face repository spent the weekend impersonating an OpenAI privacy model to deliver a Rust info-stealer to Windows users.

Three events, same week, same substrate. The processing remains entirely inside the security-tech subculture. The wider gaze hasn't found language yet for "weapons that author themselves."