For open source programs, AI coding tools are a mixed blessing

A world increasingly powered by advanced AI coding tools is supposed to be one where building software becomes cheap — or so the theory goes — leaving less room for traditional software companies to survive. One analyst report captured the idea, stating that “vibe coding will allow startups to replicate the features of complex SaaS platforms.”

That kind of thinking has fueled a wave of hand-wringing and bold predictions that software companies — and even software engineers — are on the verge of extinction.

Open source software projects, which often operate under chronic resource limitations, might seem like the most obvious winners in an era of inexpensive code, especially if AI agents can help fill the gaps. But in practice, the story is far less straightforward. The effect of AI coding tools on open source has turned out to be much more mixed than early hype suggested.

According to industry experts, AI coding tools have created nearly as many headaches as they have solved. Because these tools are so accessible, they have enabled a surge of low-quality contributions that can overwhelm open source projects. While it’s easier than ever to build new features, maintaining those features remains just as difficult — and the growing pile of code risks further fragmenting already complex software ecosystems.

The result is more complicated than a simple narrative of software abundance, suggesting that predictions of the imminent death of the software engineer may be premature.

Quality versus quantity

Across many projects, maintainers of open codebases are reporting a noticeable drop in the average quality of submissions, which many attribute to AI tools lowering the barrier to entry.

“For people who are junior to the VLC codebase, the quality of the merge requests we see is abysmal,” Jean-Baptiste Kempf, the CEO of the VideoLAN Organisation that oversees VLC, said in a recent interview.

Kempf remains optimistic about AI coding tools overall, but he believes they work best “for experienced developers.”

Blender, the open-source 3D modelling software maintained since 2002, has faced similar issues. Blender Foundation CEO Francesco Siddi said that LLM-assisted contributions often “wasted reviewers’ time and affected their motivation.” Blender is still working on a formal policy regarding AI coding tools, but Siddi said they are “neither mandated nor recommended for contributors or core developers.”

The volume of incoming merge requests has become so intense that developers are now building tools to manage the flood.

Earlier this month, developer Mitchell Hashimoto introduced a system meant to limit GitHub contributions to “vouched” users, effectively moving away from open-door contribution policies in some projects. As Hashimoto wrote in his announcement, “AI eliminated the natural barrier to entry that let OSS projects trust by default.”

A comparable issue has emerged in bug bounty programs, where external researchers can report security flaws. The open source data transfer project cURL recently paused its bug bounty program after being overwhelmed by what creator Daniel Stenberg described as “AI slop.”

“In the old days, someone actually invested a lot of time [in] the security report,” Stenberg said at a recent conference. “There was a built-in friction, but now there’s no effort at all in doing this. The floodgates are open.”

The situation is especially frustrating because open source projects are also experiencing real benefits from AI coding tools. Kempf said AI has made it much easier to build new modules for VLC — as long as an experienced developer is guiding the work.

“You can give the model the whole codebase of VLC and say, ‘I’m porting this to a new operating system,’” Kempf said. “It is useful for senior people to write new code, but it’s difficult to manage for people who don’t know what they’re doing.”

Competing priorities

For many open source projects, the deeper issue is that priorities don’t align. Large companies like Meta often emphasise shipping new products and writing new code, while open-source work tends to be driven by stability, careful maintenance, and long-term reliability.

“The problem is different from large companies to open source projects,” Kempf said. “They get promoted for writing code, not maintaining it.”

AI coding tools are also arriving during a period when software ecosystems are already highly fragmented.

Open-source investor Konstantin Vinogradov said AI tools are colliding with a long-standing reality in open-source engineering.

“On the one hand, we have an exponentially growing code base with an exponentially growing number of interdependences, and on the other hand, we have several active maintainers, which is maybe slowly growing, but definitely not keeping up,” Vinogradov said. “With AI, both parts of this equation accelerated.”

It’s a different way to frame AI’s impact on software development, with unsettling implications for the wider industry.

If engineering is defined as producing working software, AI coding tools make that easier than ever. But if engineering is more accurately described as the work of managing complexity over time, AI coding tools could end up making the job harder. At a minimum, keeping sprawling systems maintainable will require deliberate planning and sustained effort to keep complexity under control.

For Vinogradov, the result looks familiar to anyone who has spent time in open source: a growing list of tasks, and not enough experienced maintainers to handle them.

“AI does not increase the number of active, skilled maintainers,” he said. “It empowers the good ones, but all the fundamental problems just remain.”