As part of this initiative, a limited-time promotion began on March 26, 2025, allowing researchers to earn up to $13,000 for reporting Insecure Direct Object Reference (IDOR) vulnerabilities.

The bug bounty program, initially launched in April 2023 in partnership with Bugcrowd, will now offer significant rewards, including six-figure sums for exceptional findings.

OpenAI is collaborating with experts across various sectors to identify skill gaps and improve its models' ability to detect and patch vulnerabilities.

Continuous testing is being conducted to provide insights for securing AI systems against various threats, including prompt injection and other malicious manipulations.

With over 400 million weekly active users, OpenAI is committed to a proactive and transparent approach to security, addressing challenges like prompt injection attacks.

OpenAI is expanding its Cybersecurity Grant program alongside an increase in its bug bounty initiative, which aims to support research projects focused on enhancing AI security.

This expansion follows a data leak incident that affected about 1.2% of ChatGPT Plus subscribers, prompting the organization to bolster its security measures.

However, the program specifically excludes model safety issues and exploits that allow users to bypass safeguards in ChatGPT.

The organization actively monitors for malicious activities targeting its systems and works with other AI labs to enhance collective cybersecurity efforts.

Funded projects under the program have already tackled critical issues such as secure code generation and the development of autonomous cybersecurity defenses.

This new bounty initiative is part of a broader security strategy that encompasses funding for security research, ongoing adversarial testing, and collaboration with open-source communities.

OpenAI's next-generation AI projects are adopting industry-leading security practices, including zero-trust architectures and hardware-backed solutions.