Bright Security launches STAR Fuzzer with AI-written harnesses

Bright Security’s STAR Fuzzer is available now, with vendor-reported High or Critical findings in 45% of an open-source test run.

Bright Security has launched STAR Fuzzer, a mode of its Bright STAR platform at brightsec.com/fuzzer. The company says AI agents select security-critical functions and write harnesses, while a separate evolutionary search engine—not an LLM—looks for crashes, memory exhaustion and memory-safety bugs without a running application. According to Bright, the first open-source research run found High or Critical issues in 45% of projects tested, and the first month produced ten published findings in eight projects, including the Crystal standard library, htop and RIOT OS. The vendor says confirmed findings include a byte-exact payload, are replayed in a fresh container before filing, and can be delivered in a pull request. It also says a libxml2 harness reached about 13,600 regions in a 5-million-execution run.