OpenAI raises bug bounty ceiling and plans security sprint

AI · First reported 3 Nov, 02:29 · Updated 03:04

Image: Wikimedia Commons · Ysherazi · CC BY-SA 4.0

OpenAI has increased rewards for vulnerability reports, with exceptional findings eligible for up to $172,000 including bonuses. The company also plans a security sprint with Cathedral and says recent eligible submissions will receive the revised rates retroactively.

Critical vulnerabilities in core products such as ChatGPT or Codex can earn up to $30,000; exceptional findings can reach roughly $172,000 when bonuses apply.
OpenAI is preparing a security sprint with security firm Cathedral, and says 25% of its product engineers participated in the previous sprint.
The updated rates apply retroactively to eligible vulnerability reports from the previous 30 days, according to OpenAI.

Summary written by Lux AI from 2 articles

Key facts

Maximum for critical vulnerabilities: Up to $30,000 for critical flaws in core products such as ChatGPT or Codex.
Maximum including bonuses: Roughly $172,000 for exceptional findings.
Retroactive period: Eligible reports from the previous 30 days receive the difference under the new rates.
Previous sprint participation: OpenAI says 25% of its product engineers worked on finding vulnerabilities.