GPT-5.5-Cyber da OpenAI acha bug de 23 anos no OpenBSD. O ponto central é direto: OpenAI expanded Daybreak with GPT-5.5-Cyber, a security-specialized model scoring 85.6% on CyberGym versus 81.8% for the base model.
A segunda camada torna a história mais relevante: Reported findings include an old OpenBSD use-after-free issue, exploitable V8 bugs, WebKit flaws and a Firefox WebAssembly fix before Pwn2Own Berlin.
O que importa agora é execução. Patch the Planet adds human review from Trail of Bits and coordination with HackerOne so open-source maintainers are not flooded with raw model output.
Fontes: CocoLoop e reportagens públicas citadas no artigo original