A higher ceiling, with conditions
OpenAI is adding a five-tier loyalty program and time-limited campaigns that will pay more when new features launch. Researchers who reported vulnerabilities in the last 30 days will receive the difference between the old and new rates retroactively.
The $172,000 ceiling is not a direct match for every rival’s program: the comparison in the announcement is to Google’s general vulnerability rewards, while Apple and Microsoft can pay more for operating-system exploits. The headline number therefore says less about how much OpenAI will pay for any given bug than about the range of rewards it is willing to offer.
Testing alongside engineers
OpenAI is also preparing a security sprint in which specialists from Cathedral will work with OpenAI engineers to test the company’s systems. In the previous sprint, the company says, 25% of engineers working on OpenAI products took part in searching for vulnerabilities.
I think the more consequential detail is that security work is being tied to product launches, not just to a standing reward table. But the announcement gives no detail on which vulnerabilities qualify for the higher campaign rates, or how the five loyalty tiers affect payouts. Without that, researchers can see the new ceiling, but not how likely they are to reach it.
Daily AI news
Every day we pick what actually matters in AI and explain it plainly — no hype, no filler. Subscribe if you want to follow where the industry is going.
Only what matters — every day
Follow on X