Linux Foundation: maintainers see 1 in 4 PRs as AI slop

Open-source maintainers estimate that one in four incoming pull requests is low-quality "AI slop," according to a Linux Foundation report released on October 20, 2026. The same survey finds that AI tools cut the time to write code by an estimated 52%. Together, the numbers show AI speeding up contributors while adding review work for the maintainers who must check it.

The report

The report is titled "Navigating Open Source Software Development in the Generative AI Era." LF Research published it, written by Sam Boysel of the Linux Foundation and Anni Lai of the Generative AI Commons. Red Hat sponsored it, with support from Futurewei, the Open Source Security Foundation (OpenSSF) and LF AI & Data.

A pull request, or PR, is a proposed code change sent to a project for review. "AI slop" is the report's term for AI-generated changes that look plausible but are low quality.

All the figures are estimates from maintainers and contributors, not measured counts. The sources read do not give the number of respondents or when the survey ran.

The numbers

The headline figure is 25%: the share of incoming pull requests maintainers believe is AI slop. At large projects, the estimate rises to 31%, according to the Linux Foundation's announcement. Maintainer burnout from low-quality AI pull requests is the top concern at large projects, the report page says.

Finding	Share
Maintainers seeing more contributions since generative AI	59% (69% at large projects)
Maintainers who say AI lowers the barrier to entry	71%
Maintainers who disagree that AI improved community trust	53%
Contributors who review and fully test AI code before submitting	61%
Projects with a formal policy allowing AI use	35%
Projects with no policy and no plans for one	32%

The time savings vary by task. Respondents estimate 19% less time on proposal discussions and 52% less on writing code. But 21% say AI makes high-complexity tasks take longer. People who save time mostly spend it on technical debt (66%), new features (63%) and learning (58%).

Mentoring is split. 46% of maintainers say AI increases the need to mentor new contributors, while 32% say it reduces it.

What the Linux Foundation recommends

The report asks projects to stop reacting and set rules. "Open source projects need to design for a growing AI-generated contributor base, not simply react to it," said Hilary Carter, the Linux Foundation's senior vice president of research.

Its recommendations include:

• Written policies for AI-assisted contributions.
• Clearer disclosure when AI helped write a change.
• Contributors testing and documenting AI-assisted changes before they submit.
• A standard way to credit AI tools in commits.
• Treating governance and verification as core infrastructure.

Standard credit is far from settled. The Linux kernel is moving to drop its AI "Assisted-by" tag after few commits used it, the opposite of what the report recommends.

"Keeping open source secure is a shared responsibility, and maintainers should not have to carry it alone," said Adrianne Marcum, chief of staff at OpenSSF. Phoronix's Michael Larabel called the 25% figure "very concerning."

What this means for developers

The report puts numbers on a cost that contributors rarely see: the review time their AI-assisted changes create for someone else.

• Test before you submit. Only 61% of contributors say they review and fully test AI-written code first. Run the project's tests and read your diff line by line.
• Say when AI helped. Disclosure is inconsistent today. Follow the project's attribution rules, or note AI use in the pull request if there are none.
• If you maintain a project, write a policy. Only 35% of projects have one. A short CONTRIBUTING section on AI use sets expectations before the next wave of pull requests.
• Protect reviewer time. Templates, required tests and size limits on pull requests can filter low-effort submissions before a human has to read them.
• Keep mentoring. Nearly half of maintainers say AI raises the need for mentorship. Newcomers using AI still need someone to explain how the project works.