OpenAI Raises Reward to $50,000 for Universal Jailbreaks Bypassing GPT-5.6 Biosafety, Upgrades Bio Bug Bounty to Permanent Private Initiative

OpenAI has transitioned its Bio Bug Bounty program, which targets biosafety-related vulnerabilities in GPT-5.6, to a permanent operational model and significantly increased rewards.

Security Program Enhancements

Unlike previous iterations that were terminated or limited, this change allows researchers to apply to the program on an ongoing basis. In particular, the reward has been doubled to $50,000 for discovering a universal jailbreak that bypasses safety mechanisms designed to prevent bioweapon development.

Purpose and Impact

This measure aims to preemptively block the misuse of AI models for generating biological threats. Experts can participate in testing the safety limits of latest models like GPT-5.6 and reporting discovered vulnerabilities.