Flathub has drawn a firm line against AI-generated code, and the decision reaches beyond code quality alone. The platform is also responding to a rise in submission behavior that maintainers describe as increasingly rude and entitled.
Bart Piotrowski, a maintainer at Flathub, announced the policy shift on Mastodon. He said the platform now clearly does not allow AI to be used in the submission process or in the applications submitted to Flathub.
That means developers can no longer rely on code produced by an LLM if they want their app accepted into the Flatpak catalog. The restriction also covers the act of submitting the application itself, making the policy broader than a simple review of source code quality.
Piotrowski made it clear that the issue is not AI as a technology. He described AI as a potentially useful tool, both inside and outside free and open source software.
What changed, in his view, was the behavior surrounding submissions. Piotrowski said he had hoped to see more applications that reflected genuine effort from developers, but many submissions felt like little more than a prompt sent to an AI agent.
He also pointed to a sharp increase in unpleasant interactions over the past month. In some cases, he said, applicants behaved as if they were doing Flathub a favor by offering software that had been rejected.
A policy shaped by both quality and conduct
The new rule shows that Flathub is dealing with two problems at once. One is technical, centered on the growing number of applications that appear to be AI-generated with little direct developer involvement.
The other is social, tied to the way some submitters communicate with maintainers. For a platform that acts as a software gateway, that combination can make moderation and review significantly more difficult.
Flathub has not applied the ban retroactively. Apps that were already on the platform and are known to have used AI-generated code will remain available.
That approach keeps the policy focused on future submissions rather than forcing a cleanup of the existing catalog. It also avoids immediate disruption for users who already rely on apps that are currently listed.
Developers who submit new apps with code created by an LLM, however, should expect rejection. Poor conduct during the submission process is also unlikely to help a project’s chances.
Different views across the open source world
Flathub’s stance stands apart from others in the broader software community. Linus Torvalds, for example, has been reported as accepting AI-generated code as long as the code itself is good.
That contrast highlights how unsettled the question still is. Some see AI as a legitimate development aid, while others are increasingly concerned about the extra burden it places on curators and reviewers.
The pressure on maintainers is not limited to app stores and distribution platforms. Linux kernel stable release bug reporting channels have also reportedly been flooded with trivial reports from people using AI to find issues and then inflating their importance.
For end users, the new Flathub rule may not change day-to-day downloading immediately. For developers, though, the message is unambiguous: applications submitted to Flathub need to show real work and come with respectful communication.
Source: www.xda-developers.com