OpenAI has expanded its specialized Biosecurity Bug Bounty Program to include Pakistan, allowing Pakistani researchers and cybersecurity experts to win rewards of up to $50,000 (around Rs.14 million) for identifying critical security vulnerabilities in the company’s advanced AI models.
OpenAI has launched the Bug Bounty program, aimed at safeguarding the company’s most powerful and future models (such as GPT-5.6) against misuse involving biological threats.
Read Also: How to get Rs.70,000 monthly stipend through this govt-run program
OpenAI has doubled its maximum reward from previous levels to $50,000 (approximately Rs. 14 million).
Participants are tasked with discovering a “universal jailbreak”, a prompt or technique capable of consistently bypassing the models’ biosecurity safeguards. According to the program criteria, simply finding a way to solve one or two questions will not be sufficient to qualify for the top reward.
Who Can Apply:
The initiative is not open to the general public. It is aimed at professionals with expertise in AI red teaming, cybersecurity, biosecurity, or related technical fields.
Pakistan has now been added to the list of eligible countries, allowing qualified researchers to apply for participation.
How To Apply:
Interested Pakistani applicants must apply through OpenAI’s designated portal by providing their name, country of residence, institutional affiliation, and details of their relevant experience. Applicants must also have an existing ChatGPT account.
Following a review process, OpenAI will invite selected researchers to participate in the confidential program.
Successful applicants will be required to sign a Non-Disclosure Agreement (NDA) before gaining access.
The deadline for testing on the older GPT-5.5 model will remain available until 27 July 2026. After that date, the flagship program will focus exclusively on GPT-5.6.
Details of Reward:
OpenAI says researchers who successfully identify a qualifying universal jailbreak in GPT-5.5 or GPT-5.6 will receive the full $50,000 reward.
The company also announced a smaller amount for those who will identify partial security vulnerabilities.