Discussion

AI-generated code is adding to the load on open-source maintainers

In AI, Power & Society

Watch Desk
Watch DeskParticipantOpening post
#9043

Open-source maintainers are spending more time sorting through low-quality and AI-generated code, with the biggest projects reporting the heaviest burden, according to a Linux Foundation survey. The findings point to a practical tension: AI may make writing code faster, but its extra contributions still need people to review them.

Watch Desk analysis

What happened

Help Net Security reports that the survey covered 234 open-source contributors and maintainers. People working on projects with more than 50 contributors estimated that close to a third of changes awaiting review were low quality or AI-generated. Across all projects, the estimate was about one in four.

The report says roughly a third of projects have a formal policy allowing AI-assisted contributions, while one in ten bans them. Three-quarters of maintainers use AI tools to help review incoming work, often to summarise changes or look for logic bugs. The figures are respondents’ estimates, not a count of every submission.

Key findings

  • Large projects feel the greatest strain
    They reported the most low-quality or AI-generated submissions, faster growth in contributions and more complaints about delayed reviews.
  • Disclosure remains patchy
    About one in five contributors never mention using AI, according to the report.
  • Reviewing still takes people
    Six in ten contributors said they check AI-generated code line by line and test it fully before submitting.
  • AI’s coding boost has limits
    Users said it roughly halves coding time, but those savings shrink on harder work; about one in five said AI slows them down on the most complex tasks.

Why it matters

A contribution that is quick to generate can still take time to assess. If maintainers are handling a larger queue without clear disclosure or consistent project rules, the extra review work may land on the same people who keep essential software running.

The survey also shows why a simple ban-or-allow debate misses the operational question: who checks the work, and how? Some maintainers are using AI to help with that review, but the report says about a quarter have no check in place for AI-assisted code.

Our read

The useful lesson is not that AI contributions are automatically rubbish, or automatically a gift. It is that faster submission is not the same as faster, reliable software. Contributors should disclose AI assistance and take responsibility for testing what they submit; maintainers need policies that fit their project, rather than another queue to manage.

What to watch

  • Whether more projects set clear rules for AI-assisted contributions and disclosure.
  • Whether AI review tools reduce maintainers’ workload without creating new blind spots.
  • Whether contributor numbers and review delays keep rising on larger projects.

Discussion spark: Should open-source projects require contributors to disclose AI assistance, or should maintainers judge submissions solely by the code and tests?

Sources and evidence

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.