OpenAI has doubled the maximum reward offered through its specialised Bio Bug Bounty Program to $50,000 as the company intensifies efforts to identify and address safety weaknesses in its most advanced artificial intelligence models. The initiative is focused on finding vulnerabilities that could allow users to bypass built-in safeguards designed to prevent the generation of restricted biological information.
The company said the top reward will go to researchers who uncover a universal biosafety jailbreak — a prompt, method, or technique that can consistently bypass an AI model’s biological safety controls. Such discoveries would help OpenAI understand potential misuse risks and improve the reliability of its protective measures before vulnerabilities can be exploited.
The expanded programme builds on an earlier limited challenge launched for GPT-5.5 and is now being converted into a permanent safety initiative. OpenAI said testing will begin with its latest frontier model, GPT-5.6, with the objective of continuously identifying weaknesses and improving safety systems as AI capabilities continue to advance.
Also Read: Reliance Introduces Jio TeleFrame Smart Device With AI Features For Smart Homes
Unlike traditional software bug bounty programmes that usually target coding flaws or security vulnerabilities, the Bio Bug Bounty Program focuses on AI behaviour and safety performance. Researchers participating in the initiative are asked to identify ways in which models may fail to follow biological safety restrictions, allowing OpenAI to strengthen its safeguards against potential misuse.
Participation in the programme remains limited, with researchers required to apply, meet eligibility requirements, and sign a non-disclosure agreement before receiving access to testing environments. OpenAI said it may also provide smaller rewards for findings that do not qualify as universal jailbreaks but still contribute valuable information toward improving AI safety protections.
The increased bounty comes as AI companies, researchers, and policymakers place greater attention on the risks associated with increasingly powerful models. OpenAI has continued to involve external researchers in evaluating its systems, saying that independent testing plays an important role in identifying risks and developing safer AI technologies before wider deployment.
Also Read: Google Pixel 11 Pro Fold Pine Colour Variant Appears Ahead Of Official Launch