ACM Authors: AI Coding Tools Are Overloading Open Source Maintainers
By @sharedot · · 8 pages
- Programming
- Open Source
- AI Coding Tools
- Maintainers
Six ACM authors warn AI coding tools are flooding thinly funded open source projects with more code than maintainers can review.
What happened
Six authors writing for the Association for Computing Machinery's Technology Policy Council, among them Simson Garfinkel and Josiah Dykstra, argue that AI coding tools are making open source software harder to maintain and secure.
The tools write code and find security flaws quickly, but the maintainers who decide what enters a project's official release still have to judge that output themselves. AI has made writing and submitting code easy, and some of what arrives is poor.
The authors allow that quality may improve as models improve, but a human still has to assess the added volume — and most projects run on volunteer labor and sponsorships rather than reliable revenue.
Why it is surprising
The concern inverts the usual AI-productivity narrative: instead of celebrating faster code generation, the authors highlight that the bottleneck has simply moved downstream to review. Maintainer time was already a scarce resource.
At Google Summer of Code events in Bengaluru and Delhi this July, one attendee captured it in a line the Google Open Source Blog quoted: "AI can generate the slides, but the context takes nine years."
Program organizers say manually sifting through low-quality automated submissions wastes maintainer time and drains the volunteers who signed up to mentor new peers, and they are now optimizing GSoC to protect their maintainer community.
The evidence
The numbers point in the same direction from several angles.
Help Net Security reports that Google's CodeMender agent contributed 72 security fixes to open source projects between April and October 2025, including some to codebases as large as 4.5 million lines — AI can patch at scale, but every patch still needs human acceptance.
According to Forbes, the Black Duck 2026 Open Source Security and Risk Analysis Report, based on audits of nearly 1,000 commercial codebases, found the average number of open source vulnerabilities embedded in an application rose 107% year over year, driven substantially by the volume of code AI tools now let developers produce.
Ecosystem-wide data remains hard to collect because open source is governed project by project.
The stakes
Open source code sits inside phones, cars, cloud systems and AI platforms, so deferred maintenance anywhere can leave security holes in everything built on top. A Harvard Business School paper the ACM authors cite estimates firms would spend 3.5 times more on software without open source.
The funding picture is lopsided: Help Net Security reports the Linux Foundation brought in $292,217,236 in 2024, while the Apache Software Foundation, cited as an example of volunteer-run projects, brought in $2,379,402 — under 1 percent of that.
Across the ecosystem, most users never pay, a pattern known as the free-rider problem, and attackers increasingly plant malicious packages in popular repositories.
The counterargument
Not everyone reads AI-driven scrutiny as a burden.
A Forbes Technology Council post by OpenVPN CEO Francis Dinha argues AI turns open source inspectability into verification at scale: AI can audit a codebase and surface critical vulnerabilities in minutes, a task that used to take researchers weeks, and defenders can scan the same public code at the same speed as attackers.
He points to GitHub's Secure Open Source Fund, which reportedly resolved over 4,000 CodeQL security alerts and blocked more than 100 exposed secrets in six months.
But that help still flows through maintainers who must triage, accept and release — the same scarce human capacity the ACM paper says AI is exhausting.
What comes next
The ACM authors want projects to put more resources into documentation, packaging, fundraising and gathering requirements, because deciding what a project needs and building community agreement still falls to people.
They also flag that despite US and EU mandates, the vast majority of open source applications do not generate or ship a software bill of materials, leaving organizations blind to which dependencies are maintained, funded or abandoned — a gap they call perhaps institutions' largest challenge.
The GSoC program, for its part, says it will tune its rules to shield maintainers from automated low-quality submissions so mentors can keep building the pipeline that turns first-time contributors into long-term stewards.
Sources
- helpnetsecurity.com › AI is adding to the review load on open-source projects, many of them thinly funded
- opensource.googleblog.com › Reconnecting with the heart of open source: Highlights from our 2026 GSoC India tour
- forbes.com › Open Source In The Age Of AI: Why Transparency Is Your Best Security Control