FT: AI Coding Boom Is Overwhelming Open-Source Maintainers

The Financial Times has a piece out arguing that the AI coding boom is quietly draining the open-source ecosystem that most of the boom actually runs on. The framing is a tragedy of the commons: a developer using a coding assistant gets faster, the maintainer on the other end of the pull request gets a larger pile of superficially plausible, low-context contributions to triage, and nobody is paying them for that extra work.

The evidence sits on the maintainer side, and it is now concrete. Daniel Stenberg shut down cURL's six-year bug bounty program earlier this year after roughly $86,000 in payouts, with valid submissions dropping to about 5% as AI-generated reports climbed. Mitchell Hashimoto's Ghostty banned AI-generated code submitted without approval, with Hashimoto insisting the line 'is not an anti-AI stance' but 'an anti-idiot stance.' Steve Ruiz's tldraw went further and now auto-closes all external pull requests. RedMonk analyst Kate Holterhoff has been calling the pattern 'AI Slopageddon' in her own writing.

The research the reporting leans on is a Central European University and Kiel Institute for the World Economy paper titled 'Vibe Coding Kills Open Source'. It models what happens when AI agents assemble applications by pulling in open-source packages without any of the follow-on engagement — documentation reads, issue reports, upstream contributions — that historically compensated maintainers through visibility, sponsorships and consulting work. Productivity rises for the AI user; maintainer incentives fall, and the pipeline of serious new contributors thins alongside them. As proof points, coverage cites Stack Overflow activity dropping about 25% within six months of ChatGPT's launch, and Tailwind CSS documentation traffic falling 40% while revenue declined 80%.

The honest caveat is that the strongest numbers here are attention-flow proxies, not clean measures of ecosystem health, and the FT is stitching together maintainer anecdotes plus a single working paper into a big-picture story that will take longer than one quarter to actually confirm. What the reporting does not give you is a platform-level answer: whether GitHub, GitLab and the big foundations build slop filters and fund the maintainers they depend on before this pattern locks in.

If you ship on top of open source, meaning almost everyone, the actionable read is unglamorous. Assume the maintainers behind your critical dependencies are getting louder about this, budget sponsorship into infrastructure spend, and expect more projects to follow tldraw's lead and simply close the door to outside PRs.