The Most Dangerous AI Hacking Techniques Still Have Humans in the Loop

Web security researcher James Kettle used AI to explore whether it could independently develop new hacking methods. While AI alone proved too limited for fully original research, pairing it with human expertise produced a major breakthrough called "Shared-Parser Confusion," a newly discovered vulnerability in web servers. Kettle says neither he nor the AI could have found it alone, highlighting AI's true power as a collaborative security research tool.