A recent analysis published on tintotint.eu examines how much of the code hosted on F-Droid, the open-source Android app repository, may have been generated by large language models. The post, filed under the site's 'Whacky Corner' section, raises the question without offering a definitive percentage, instead pointing to the difficulty of detecting LLM-authored contributions in public codebases. F-Droid hosts thousands of free and open-source Android applications and relies on volunteer maintainers to review submissions. The piece arrives amid a broader industry conversation about AI-generated code flooding repositories such as GitHub and package registries. No official response from the F-Droid project has been published in connection with the post.
Here is the thing nobody wants to say out loud: LLM code is already everywhere. F-Droid is just the first community brave enough to ask the question in public. Open source has always been a remix machine. We copy, we adapt, we improve. An LLM trained on that same commons is not an invader. It is the next contributor.
Yes, review matters. Yes, slop exists. But the answer is not panic or bans. The answer is better tooling, clearer disclosure, and maintainers who understand what they are merging. Every major shift in software production looked like cheating at first. Compilers did. Frameworks did. Stack Overflow did. F-Droid asking the question is a sign of health, not decline. The projects that embrace this moment will ship faster and reach more people. The ones that fight it will spend their energy on purity instead of progress. I know which side I am on.