Phoenix Security has launched Exploit Hunt, an AI red-team capability that works from a live threat model and reports a vulnerability only after generating and validating a runnable proof-of-concept exploit. Announced Aug. 5 and timed to Black Hat USA 2026, Exploit Hunt is available now in Phoenix Purple. Exploit Hunt then uses three separate personas. VERIFY attempts to turn surviving findings into a runnable exploit and validation document. In testing on a vulnerable-application, multi-repository corpus, Phoenix said the system turned three target files into nine confirmed exploits involving JWT handling, cross-site scripting and annotation injection.