A critical vulnerability in a core product such as ChatGPT, Codex or the API Platform will now earn up to $30,000, OpenAI said on Nov. 2 in its announcement. Exceptional critical findings on those targets stay eligible for up to $100,000, and now up to $172,000 with bonuses.

What changed in the program

OpenAI says it is "more than quadrupling the midpoint of our reward ranges for medium, high, and critical findings on average." A companion post on its next security sprint uses the phrase "more than quadrupling our rewards for critical findings."

Old and new, in OpenAI's examples

Figures here come from OpenAI's two Nov. 2 posts. Its four examples show how the ranges and multipliers combine; they are illustrative, not real payouts.

  • Medium, ChatGPT core: $750 under the old ranges, $2,000 now. With a 1.15x quality bonus and a 1.05x loyalty tier, $2,415.
  • High, Prism non-core (OpenAI's label): $1,000 before, $2,500 now. With 1.15x and 1.10x, $3,162.50.
  • Critical, ChatGPT core one-click account takeover: $6,500 before, $30,000 now. With a 1.15x quality bonus, a 1.30x priority campaign and a 1.15x loyalty tier, about $51,577.
  • Exceptional, ChatGPT core: $100,000 before, $171,925 now with bonuses.

Bonuses on top of the base award

OpenAI's examples combine up to three multipliers on the base award:

  • Quality bonus: a multiplier for high-quality reports with a reliable proof of concept.
  • Loyalty tiers: five tiers. The examples name Bronze at 1.05x, Gold at 1.10x and "31337" at 1.15x.
  • Priority campaigns: short-term bonus pools for newly launched features. OpenAI's example is 1.30x.

Recent awards get a top-up

Reports already submitted without a bounty awarded get the new ranges. For bounties awarded in the 30 days before the announcement, OpenAI says it will pay the difference if the finding would have earned more under the new ranges, counting the base award and loyalty awards. OpenAI's security team will also be at PROMPT2PWN in Oakland on Nov. 20-21.

Who is affected

This matters to security researchers who report to OpenAI. The announcement is about researcher rewards.

Why it matters

The $30,000 and $172,000 figures apply to core products: ChatGPT, Codex and the API Platform. OpenAI's own non-core example, a high-severity Prism finding, has a $2,500 base. The top-up covers only unpaid reports and bounties awarded in the 30 days before the announcement, counting base plus loyalty awards.

Related

For another ChatGPT security story, see ChatGPT shared link vulnerability let attacker code run in a victim's account. More is in AI & Apps.