OpenAI doubles top reward to $50,000 for GPT-5.6 jailbreak researchers
OpenAI just doubled its top reward to $50,000 for eligible researchers who can find ways to jailbreak its newest AI model, GPT-5.6.
The Bio Bug Bounty Program is all about spotting tricks that could make the AI break safety rules and generate banned biology-related content.
What started as a short-term challenge for GPT-5.5 is now a permanent part of OpenAI's safety efforts.
OpenAI's bio bug bounty requires NDAs
This program isn't your usual software bug hunt. It's focused on keeping AI from crossing dangerous lines in biology.
To join, researchers have to apply and sign NDAs before testing the model. Smaller rewards are given for helpful feedback that makes the system safer.
OpenAI's move fits into a bigger push across tech to stress-test powerful AIs as they get smarter and more widely used around the world.