The numbers are finally in, and they paint a fascinating picture of human behavior in modern software development. A comprehensive study of the Linux app repository Flathub reveals that 73% of applications built using AI generators were completely abandoned within months of their initial release.
The data directly validates a highly debated policy implemented just two months ago. In May 2026, Flathub maintainer Bart Piotrowski issued a strict ban on AI-assisted applications. He rewrote the platform’s guidelines to prohibit artificial intelligence tools in both the application code and the GitHub pull request submission process.
The 120-Repository Reality Check
Linux developer Evangelos Paterakis ran the numbers on 120 unique software repositories flagged by Flathub reviewers with an internal “AI Slop” tag. This designation marks apps that rely heavily on generated code. Out of those 120 projects, only 32 showed any ongoing maintenance.
The remaining 88 repositories sat entirely dead. They received zero significant updates over the past two to three months. In several instances, the original submitters deleted the source code entirely once the initial fun wore off, according to data published on the platform’s abandonment rates.
Reviewers Pushed to the Limit
Piotrowski noted that the May ban was a direct result of user behavior rather than pure code quality. A flood of low-effort, “vibe-coded” software quickly overwhelmed the platform’s volunteer-run review queue. When reviewers requested standard manual fixes to these generated apps, they frequently encountered hostile arguments from developers who refused to write the corrections themselves.
How Generated Code is Fracturing Open Source
The friction on Flathub mirrors growing frustration across the broader open-source community. Major projects are taking distinct, often opposing, stances on generated contributions. The GNOME desktop environment recently updated its own extension guidelines to actively block generated code. Maintainers for the Linux kernel also report receiving a high volume of trivial, AI-generated bug reports that waste development time.
But not every open-source leader agrees with the strict bans. Linux creator Linus Torvalds takes a completely different approach. He recently stated he will “loudly ignore” critics of AI code, maintaining that as long as the submitted contributions are functionally sound, the origin of the code simply does not matter.