OpenAI raises rewards for reporting critical ChatGPT and Codex flaws
OpenAI has revised the payments offered to researchers who privately report security flaws. Critical weaknesses in core ChatGPT and Codex functions have a higher reward ceiling, while exceptional findings and bonuses can produce larger payouts. A new loyalty structure and retrospective adjustments also change how eligible contributions are rewarded.
Artificial Intelligence··Evening
Higher ceilings for critical flaws
OpenAI has increased the rewards available to outside researchers who report vulnerabilities in its artificial intelligence products. Critical weaknesses affecting central ChatGPT and Codex functions can earn up to 30,000 dollars. Exceptional submissions qualify for a larger reward before additional bonuses. With those bonuses included, the combined ceiling reaches roughly 172,000 dollars. These amounts apply to different categories and conditions within the disclosure program.[1], [2]
Contribution history enters the calculation
The revised program introduces five loyalty tiers reflecting researchers’ contributions. The quality of a report can also attract an extra payment. The time-limited campaigns offer further incentives connected to new product features. A researcher’s final reward therefore depends on the finding and the applicable additions, rather than a single payment assigned to every submitted vulnerability. The changes aim to encourage useful reports through OpenAI’s designated security disclosure process.[1]
Recent submissions receive back payments
OpenAI is applying the revised rates to qualifying reports submitted during the preceding 30 days. Researchers covered by that adjustment will receive the difference between the previous award and the new amount. The retrospective payment extends the change to work already delivered within that period. The new ceilings concern the reward rules for disclosure; the announcement does not report that the listed weaknesses were exploited in an attack.[1]
