OpenAI is transforming its GPT-5.5 Bio Bug Bounty into an ongoing private program, the Open AI Bio Bounty Program. This initiative targets universal jailbreaks that can bypass safety measures for AI models in biological applications, starting with GPT-5.6. The program aims to proactively identify and mitigate risks associated with advanced AI capabilities in biology.

Previously offering $25,000, the reward for a successful universal jailbreak has now doubled to $50,000. This increased incentive applies to both GPT-5.6 and GPT-5.5. While the GPT-5.5 testing concludes on July 27, 2026, GPT-5.6 will remain the focus thereafter. Smaller rewards may be issued at OpenAI's discretion for partial successes.

Researchers can apply through a rolling application process, requiring a brief submission of their name, affiliation, and experience. Accepted participants must have an existing ChatGPT account and will sign a non-disclosure agreement. This program represents a structured approach to externally validate and strengthen the biosafety protocols of OpenAI's frontier models.