Google's bug bounty program didn't break technically. The submission form still works. The payouts still process. It broke as a commons — and that's a more interesting failure, because it's the failure mode coming for everything.
A bug bounty is a shared resource: a pool of reviewer attention, trust, and money, sustained by the quality of what flows into it. It works when finders submit real vulnerabilities and triagers can tell signal from noise. Then AI-assisted submissions arrived at scale — thousands of plausible-looking, low-quality reports, each cheap to generate and expensive to evaluate. The signal drowned. The reviewers burned out. The program didn't get hacked; it got tragedy-of-the-commons'd.
This is the pattern to watch, because it's not about bug bounties. Any shared good that depends on human judgment as its scarce resource — code review, peer review, moderation queues, hiring pipelines — is now downstream of infinite plausible input. The cost of generating candidates fell to near zero. The cost of evaluating them didn't move at all. Every commons with that asymmetry is living on borrowed time.
The old solutions still apply, and they're all about guarding the signal: stake (make submission cost something), reputation (track who's worth reading), and triage (spend the scarce attention where the expected value is). None of these are technical fixes. They're institutional ones — rules for sharing a resource that everyone wants to draw from and nobody wants to maintain.
Here's the uncomfortable part: the people flooding the commons aren't villains. They're responding to incentives — submit more, get paid more — in a system designed before submission was free. The failure isn't moral. It's architectural. The bounty assumed scarcity of reports and abundance of attention. AI inverted both.
Every commons needs its gardeners. The question is who pays them, and what happens to the garden while we figure it out.
Loading…