AI-Assisted HTTP Terminator Finds Novel HTTP Desync Techniques and Apache Zero-Day

An AI system named HTTP Terminator, developed by James Kettle, has discovered novel HTTP desynchronization techniques by analyzing 30,000 candidate vectors. The research uncovered vulnerabilities in various sectors, including finance and government, and introduced new methods like the 'dangling-byte' technique to improve response queue poisoning attacks. Additionally, a separate human-guided discovery led to a zero-day vulnerability in Apache Traffic Server, now patched and tracked as CVE-2026-63078.