Can you break ChatGPT? OpenAI offers up to $50,000 for universal jailbreaks

OpenAI has reportedly doubled the maximum award to $50,000 for researchers who can develop a universal jailbreak bypassing its biological safeguards. This is part of an ongoing private program that features GPT-5.6 while keeping GPT-5.5 in scope through July 27, 2026.

This move is designed to reward researchers who can find a reusable way to breach the model’s biological safeguards. 

The company has already announced plans on July 9 that researchers may earn up to $50,000 for a universal jailbreak that passes a microbial challenge.

OpenAI has launched its most capable model, GPT-5.6 on Thursday. The primary reason behind the delay last month at the request of the US government was security challenges regarding how easily powerful AI systems could be misapplied.

OpenAI has restricted access to the model to a selective group of strategic alliances whose details were shared by the authorities. 

US President Donald Trump has signed a presidential directive establishing a voluntary framework for AI developers to offer covered frontier models to the US government for up to 30 days before releasing them to trusted partners.

OpenAI Bio Bounty program

The new OpenAI Bio Bounty program originally covers GPT-5.6 while keeping GPT-5.5 in scope while providing specialists an ongoing way to use protections so that enterprises can safely depend on advanced AI models in sensitive research environments.

The original GPT-5.5 bounty launched on April 23, 2026. At the time when researchers had to use one prompt to answer five biosafety questions from a clean Codex Desktop session without flagging the content. OpenAI has not confirmed whether GPT-5.6 uses similar questions for testing purposes.

The company has not yet disclosed successful submissions or related mitigations. As AI systems gain access to files, email, code and business applications, audit trails become more significant for monitoring, approvals and human review. 

Nonetheless, only public reporting and strategic repositioning will shape outside scrutiny of the entire program.