On October 20, 2026, the Linux Foundation announced a Linux Foundation Research estimate that 25% of incoming open-source pull requests are low-quality, AI-generated contributions. The figure reflects estimates from 111 survey respondents, not an audit of submitted code, and the report covers open-source projects broadly rather than the Linux kernel alone.
A pull request (PR) is a proposed code change submitted for review. The report’s chart describes the estimated share as “low-quality and/or AI-generated,” while its narrative defines “AI slop” as low-quality contributions generated with AI. That wording matters: the headline number is a survey estimate, not a measured rate across open-source repositories.
What the 25% estimate represents
Linux Foundation Research conducted a 45-question web survey from April through June 2026. The overall survey included 234 qualified respondents; 111 respondents supplied estimates for the PR figure. The report presents the 25% as a mean estimate of incoming contributions.
How estimates differ by project size
The reported mean estimate was higher for large projects than for small or medium ones. Linux Foundation Research defines project size by contributor count:
| Project size | Contributors | Estimated share |
| Small | Fewer than 10 | 23% |
| Medium | 11–50 | 20% |
| Large | More than 50 | 31% |
These are respondent estimates for each project-size group, not audited shares of actual pull requests. The 31% figure gives a different picture from the 25% overall estimate, especially for maintainers of larger projects.
AI time savings depend on the task
Survey respondents estimated an average 52% time reduction for code authoring and 19% for proposal discussions. For high-complexity tasks, 64% said generative AI saved time, 21% said it required more time, and the remaining respondents reported no change. The results describe different tasks and self-reported estimates, not a uniform productivity gain.
Project policies and the pressure on maintainers
Among 119 respondents asked about project policies, 35% said their projects formally permitted generative-AI contributions. Another 32% had no policy and no plans for one, 23% planned to establish a policy, and 10% did not permit AI-assisted contributions.
| Reported project policy | Share |
| Formally permits generative-AI contributions | 35% |
| No policy and none planned | 32% |
| Plans to establish a policy | 23% |
| Prohibits AI-assisted contributions | 10% |
The survey also found that 59% of maintainers saw open-source contribution volume rise after generative AI emerged, a figure that reached 69% for large projects. On community effects, 71% agreed that generative AI lowers barriers to contributing, while 53% disagreed that it had improved community trust. Nearly half, 46%, said AI increased the need to mentor new contributors; 32% said it decreased that need.
Linux Foundation Research recommends written policies that clarify AI-use disclosure and attribution, contributor testing and documentation before submission, and continued human mentorship. It also suggests using AI tools for maintenance tasks where they help. These are recommendations in the report, alongside survey estimates of how project policies and contribution patterns are changing.
