OpenAI launches Bio Bug Bounty for GPT‑5.5: seeking universal jailbreak for biosecurity

Friends, sharing news from OpenAI’s ecosystem: a Bio Bug Bounty for GPT‑5.5 has been launched.

Brief:
- What: a contest for vetted AI/red‑teaming and biosecurity researchers to find a universal jailbreak that answers five biosecurity prompts from a clean chat.
- Where: GPT‑5.5 on Codex Desktop; participation by invitation and application, all communications under NDA.
- Prizes & timeline: $25,000 for the first full bypass; submissions open until 22 June 2026; testing runs 28 Apr–27 Jul 2026.

Why it matters: the program evaluates real attack vectors and strengthens the model’s resilience to bio risks.

Should we monitor results and consider collaborating with participants?

#OpenAI #biosecurity #cybersecurity #AI