Executive Summary
OpenAI is transforming its temporary GPT-5.5 Bio Bug Bounty into an ongoing, private initiative named the OpenAI Bio Bounty Program. The program incentivizes researchers to discover "universal jailbreaks" that could bypass safety protocols related to biological risks in the company's frontier AI models. To encourage participation, OpenAI has doubled the maximum reward for a successful universal jailbreak from $25,000 to $50,000, with the new program focusing on models from GPT-5.6 onward.
Key Takeaways
* Program Evolution: The one-time GPT-5.5 Bio Bug Bounty is now a permanent, private program called the OpenAI Bio Bounty Program.
* Increased Reward: The maximum payout for discovering a universal jailbreak related to biorisks has been doubled to $50,000. Partial awards may be given for less severe findings.
* Updated Model Scope: The program will now focus on current and future frontier models, starting with GPT-5.6. The bounty for GPT-5.5 will conclude on July 27, 2026.
* Participation: The program is open to selected researchers via a rolling application process, and accepted applicants must sign an NDA.
Strategic Importance
This move signals OpenAI's heightened commitment to proactively identifying and mitigating catastrophic biological risks in its most advanced models. By formalizing the program and increasing the financial incentive, the company aims to strengthen its safety safeguards and build trust with regulators and the public.