Last week, a former OpenStack and Kubernetes contributor asked a simple question on Hacker News: with AI coding tools advancing rapidly, is there still room for humans in open source? The thread drew two comments and a few hundred views, but the question underneath it is one the entire software ecosystem is now grappling with.

A Deluge of Low-Effort Contributions

Open source projects are drowning in AI-generated pull requests. Researchers have documented what they call an "AI-DDoS" effect: plausible but low-quality contributions flooding project inboxes faster than maintainers can triage them. At the cURL project, a six-year bug bounty program was shut down in January 2026 after the share of valid submissions collapsed to roughly 5% by mid-2025. The volume was simply unsustainable.

Across 294 widely used projects, pull request volume has climbed while merge rates and issue completion rates have fallen since 2025. The hardest-hit contributors are one-time submitters, many of whom appear to be using AI agents to generate drive-by contributions and then disappearing. GitHub itself has described 2025 as an "Eternal September" for open source, where sheer contribution volume overwhelms human attention and erodes trust between contributors and maintainers.

Who Is Actually Writing the Code

The numbers behind this shift are striking. Between October 2025 and May 2026, the share of pull requests to critical open source projects containing AI agent co-authors grew from 2% to 10% -- a sixfold increase in six months. Human-authored contributions rose from roughly 300,000 to 400,000 in the same period, likely supported by AI assistants, but showed a decline for two consecutive months starting in March 2026.

At this pace, agentic AI will outnumber human contributors in pull request volume within a year. Meanwhile, the share of rejected versus accepted AI-assisted contributions has decreased, suggesting these pull requests are not necessarily low-quality. But they still require manual review, and many critical codebases are maintained by one or two volunteers who are burning out under the load.

Newcomers Without Ownership

An empirical study of open source review practices found that over 67.5% of AI-co-authored pull requests originate from contributors who have no prior code ownership in the project. Most of these contributors have never worked on the codebase before. Yet 86.9% of projects still lack any guidelines governing AI agent usage.

AI-assisted pull requests are also being merged significantly faster, with minimal reviewer feedback, compared to human-only submissions. This creates a paradox: the contributions that arrive with the least human context and the shallowest understanding of the codebase are the ones making it into the code fastest.

Projects Are Building Gates

Maintainers are responding. The Ghostty project now requires contributors to disclose AI tool usage, demonstrate human understanding of submitted code, and faces sanctions for low-effort AI-generated work. Other projects have added stronger review gates or stepped back from accepting outside contributions entirely.

The cURL project's decision to close its bug bounty illustrates the broader risk: when communities close their doors to survive the flood, they cut off the contributor pipeline that renews maintainer labor. Open projects risk becoming closed systems that slowly atrophy through attrition, weakening the critical software supply chains that depend on them.

The Contributor as a Taste Maker

Despite the noise, experienced contributors report that the core experience of working in open source communities has not fundamentally changed. A commenter on the Hacker News thread summed it up plainly: people have had to deal with a deluge of AI slop, but hang around a community long enough and it still feels the same, with its good and its bad.

The Linux Foundation's 2026 report on open source and AI frames the evolving role of the human contributor as a "taste maker" -- someone who defines the problem, sets the standards, and applies the final human-owned stamp of quality to AI-generated output. That is a meaningful shift from writing code to directing its creation. It rewards judgment, domain knowledge, and design sense over raw typing speed.

What It Means for Would-Be Contributors

For someone with a background in projects like OpenStack and Kubernetes looking to return to open source, the landscape is more complex but not closed. The barrier to generating code has dropped to near zero. The barrier to generating useful code that a community will accept and maintain has not.

Projects increasingly value contributors who understand the system, can review AI output critically, and engage with the community beyond submitting patches. The most in-demand skill is no longer writing code from scratch. It is knowing what code should exist, what belongs in the project, and what does not.

The ecosystem is still growing. GitHub's Octoverse 2025 report documented over 4.3 million AI-related repositories and a 178% year-over-year jump in LLM-focused projects. Open source powers more than 90% of modern software applications. The question is not whether open source contribution is relevant. It is whether contributors are prepared to evolve alongside the tools that have changed the nature of the work.