#

Key takeaways

• OpenAI increased reward ranges for medium, high, and critical findings.

• A critical issue in core products can now pay up to $30,000.

• Exceptional critical findings can still reach $100,000 before bonuses.

• The program now includes quality, loyalty, and priority campaign multipliers.

• New ranges apply to reports submitted but not yet awarded.

OpenAI updates its Security Bug Bounty program

OpenAI announced changes to its Security Bug Bounty program on November 2, 2026. The update raises the midpoint of reward ranges for medium, high, and critical findings by more than four times on average. It also adds new bonus structures that can increase payouts for certain reports.

The company says a critical vulnerability in a core product such as ChatGPT, Codex, or the API Platform can now receive up to $30,000. Exceptional critical findings remain eligible for up to $100,000 before bonuses. OpenAI says that amount can reach approximately $172,000 including bonuses.

How the new bonuses work

OpenAI added multipliers for high-quality reports, loyalty, and priority campaigns. The quality bonus is for reports with reliable proofs of concept. Those reports should isolate a weakness, explain exploitability, and demonstrate impact. That makes the reporting standard more specific and more structured.

The loyalty program has five tiers. It rewards repeated collaboration with the company. OpenAI also introduced short-term priority campaigns. These campaigns use dedicated bonus pools for newly launched features.

The source includes payout calculations to show how bonuses can combine. These examples are illustrative scenarios and historical awards. They are not disclosures of newly discovered vulnerabilities. That distinction matters because the examples should not be read as evidence of a current flaw in any product.

Timing and payout adjustments

OpenAI says the new ranges apply to reports that have been submitted but not yet awarded. It also says that for bounties awarded within the thirty days before the announcement, it will pay the applicable difference under the new base and loyalty ranges. This creates a retroactive adjustment window for recent awards.

The announcement also describes participation in PROMPT2PWN in Oakland on November 20–21. Researchers must follow the actual program scope and terms. The announcement does not grant permission to test unrelated systems. That means the program boundaries still control what researchers can and cannot assess.

What this means for researchers

The update signals a stronger incentive structure for security reporting. Higher base ranges, plus bonuses, can make the program more attractive to researchers who can produce clear, reproducible findings. The emphasis on proof of concept quality also suggests that report clarity now plays a larger role in payout outcomes.

The loyalty tiers may also encourage ongoing participation. Researchers who work with the program repeatedly can move through the five-tier structure. Priority campaigns add another layer, since they direct bonus pools toward newly launched features.

For researchers, the practical takeaway is simple. Strong reports, clear evidence, and attention to program scope matter more than ever. The announcement ties reward size to report quality and participation history, while keeping the program rules in place.

Editorial note

The payout figures in the source are examples and historical awards. They should not be treated as claims about active vulnerabilities. The announcement describes reward policy changes, not a security incident.