Why Open Source Is Drawing the Line on AI Slop

The open-source game engine Godot recently added one rule to its contribution guidelines: "contributions made entirely by AI are prohibited." It is not an isolated case. Roughly 96% of the world's software relies on open source, yet about 60% of the maintainers holding it together are unpaid volunteers. In 2026, these open source maintainers started drawing defensive lines against a flood of low-effort, machine-generated pull requests — the vibe coding output they label AI slop.

Godot's AI Contribution Ban

Godot's guidelines now discourage AI use and ban pull requests written entirely by AI. Anyone who used AI for any part of a change must disclose where, and be ready to explain the code to maintainers. The only exceptions are single-line autocompletion and translation tools. The full terms sit in the project's contribution guidelines.

Rémi Verschelde, a core maintainer of nearly a decade, explained why on Bluesky: vetting AI slop is "draining and demoralizing." Maintainers must now second-guess every contribution from a new developer — was it written by a human, and does the author even understand it?

His only practical fix was more funding to hire maintainers who can filter the influx. Running AI to detect AI-generated code, he added, would be "horribly ironic." At the time, Godot had 4,681 open pull requests awaiting review.

curl Suspends Bug Bounties, Jazzband Shuts Down

Projects without funds for more staff took sharper measures. Daniel Stenberg, curl's lead developer, said the AI slop flooding his inbox — low-quality, AI-generated bug reports — was effectively a "denial-of-service" attack. He shut down curl's bug bounty in January 2026, reasoning that the reward only drew automated spam. Valid vulnerability reports had already fallen from 15% to 5%.

Jazzband, a collaborative collective within the Python ecosystem, chose to disband entirely. Lead maintainer Jannis Leidel, who also serves as the chair of the Python Software Foundation, wrote that the relentless "flood of AI-generated spam PRs and issues" had made the project unsustainable.

The economics of AI slop are stark. One developer estimated that reviewing and fixing a pull request takes 12 times longer than generating it. The vibe coding that takes thirty seconds to produce can need days to verify — an unsustainable load for the unpaid maintainers who keep these projects alive.

GitHub Battles AI Slop as Trust Breaks Down

The strain forced the platform itself to act. In late January, GitHub product manager Camilla Moraes opened a community discussion on how the surge in AI slop had become an operational crisis for maintainers. Proposed fixes include disabling pull requests entirely, restricting them to established collaborators, and adding metadata to flag AI use.

On the front lines it is worse. One core developer said only about 10% of pull requests churned out by vibe coding clear basic quality bars. The PlayStation 3 emulator RPCS3 has threatened bans for undisclosed AI code. A GoCD maintainer warned that autonomous coding agents like OpenClaw — the subject of our OpenClaw profile — will make it worse, after one documentation pull request proved "plausible nonsense" only after long review.

The deepest damage is to trust. As a Microsoft engineer put it, reviewers can no longer assume a contributor wrote or even understands their submission. AI-generated pull requests can look structurally sound yet hide logical errors or security holes, and line-by-line review cannot scale to the volume of machine-written code.

The Real Target: Accountability, Not the Tool

The community is not unified in banning AI. Of 63 open-source AI policies catalogued by RedMonk analyst Kate Holterhoff, only 14 ban AI-generated contributions outright; 12 remain undecided, while most permit AI if its use is disclosed. Even the strictest ban turns on one question — does the contribution come with real understanding?

One pattern holds, though: the closer a project sits to critical infrastructure, the stricter its open source AI policy. The line is drawn not against AI the tool, but against vibe coding thrown over the wall without accountability. It echoes the principle Godot set long before any AI ban — submit only code you understand and can explain.