Session

The Bug Arms Race: Surviving AI-Generated Bug Reports in Open Source

What happens to an open-source project when both sides get AI?

At ArcadeDB, a small team adopted an AI-native workflow, and our output jumped 28×. Visibility followed, and so did a 13× jump in AI-generated bug reports. Finding bugs had become free. Verifying them had not.

This is a data-driven account of that arms race, taken from the public ArcadeDB repository. You'll see what broke: triage bandwidth, maintainer focus, and the signal-to-noise ratio in the issue tracker. You'll also see what we built to survive: intake policies for AI-generated reports, a test net strong enough to confirm or reject a claim automatically, and an agent pipeline that uses AI to fight AI.

For maintainers, tech leads and engineers adopting AI-assisted development who want to scale output without drowning in it.

Roberto Franchini

Principal Software Engineer @ ArcadeData

Turin, Italy

Actions

Please note that Sessionize is not responsible for the accuracy or validity of the data provided by speakers. If you suspect this profile to be fake or spam, please let us know.

Jump to top