Saved report copy
Recorded
AI makes years-old PHP bugs easier to exploit
Security researchers at Calif, OpenAI, and Wiz used AI tools to uncover years-old vulnerabilities in the PHP scripting language, collectively called PHP2Shell, and criminal hackers quickly tried to exploit them, with millions of attempts logged within hours of disclosure. The flaws were known for years but were considered too complex for humans to exploit, researchers said. The PHP Foundation has not confirmed any successful attacks, but the episode highlights how AI is broadening which bugs are practically exploitable and adding pressure on security teams to keep up.