AI-Assisted Research Uncovers Critical PHP Security Flaws

Researchers used advanced artificial intelligence models to help identify serious vulnerabilities in PHP, the programming language powering a large proportion of websites. According to The Rundown AI, security specialists at Calif and Wiz independently investigated the flaws, with models including GPT-5.6 Sol helping with the discovery. Maintainers worked on patches as governments and businesses prepared urgent responses. Some public websites were temporarily taken offline while administrators assessed exposure, and security companies reported early attempts to exploit the weaknesses. The incident shows how AI can support defensive security research, while also raising concerns that attackers may use similar capabilities to discover vulnerabilities. The report emphasises the importance of prompt patching and coordinated disclosure across critical open-source infrastructure used throughout the internet.

The Rundown AI, Zach Mink – AI finds a security hole in 70% of the web – 22nd October 2026