OpenAI on November 2, 2026 raised the payouts in its Security Bug Bounty program, the channel through which outside researchers report vulnerabilities in its software. The company says it is more than quadrupling, on average, the midpoint of its reward ranges for medium, high and critical findings, and adding new multipliers meant to pay more for better reports and for researchers who keep coming back.
For core products such as ChatGPT, Codex and the API Platform, a critical vulnerability can now earn up to $30,000. Exceptional critical findings on those targets remain eligible for up to $100,000, the ceiling OpenAI set in March 2025, but with the new bonuses a single award can now reach about $172,000. When OpenAI opened the program in April 2023, BleepingComputer reported at the time, the top reward was $20,000.
Three kinds of bonus sit on top of the base award. A quality multiplier rewards reports with a reliable proof of concept that isolates the bug, shows when it can be exploited and demonstrates its impact. A five-tier loyalty scheme scales bonuses for researchers who report consistently. Priority campaigns will attach dedicated bonus pools to newly launched features for short periods, so that early testers of new products earn more.
OpenAI's post illustrates how the pieces combine, while stressing that the scenarios and historical awards are illustrative. A medium-severity ChatGPT bug that once paid $750, letting a read-only member upload files to a Project, would now pay $2,415 after quality and bronze-tier bonuses. A one-click ChatGPT account takeover previously rewarded with $6,500 would pay $51,577 with quality, priority-campaign and top-tier loyalty multipliers. The top example, a significant compromise of a security boundary such as viewing other users' chats, works out at $171,925.
The change is partly retroactive. Reports already submitted but not yet rewarded will be paid under the new ranges, and researchers who received a bounty in the 30 days before the announcement will get the difference if their finding would have earned more. OpenAI also says its security team will be at the PROMPT2PWN event in Oakland on November 20 and 21, where researchers can test targets in person.
The bounty update came with a second announcement the same day. OpenAI says it is planning its next internal security sprint and will bring in Cathedral Systems, whose researchers will use OpenAI's models alongside its own teams to probe its defenses. The company says that in its last sprint it dedicated 25% of its product engineers to finding and fixing weaknesses, using Codex to investigate findings, reproduce vulnerabilities in isolated environments and prepare patches for review. It describes outside bounty hunters as a complement to that internal, model-driven work, which it calls Defense Factory.
The figures are maximums and multipliers rather than guaranteed payouts, and OpenAI has not said how many reports it receives or how much it has paid out to date. Model safety issues such as jailbreaks have historically been out of scope for this program, according to BleepingComputer's 2025 report, so the new money targets conventional software vulnerabilities in OpenAI's products rather than model behavior.