A recent Hacker News thread with 68 points and 77 comments asks how much of F-Droid now contains LLM-generated apps. The discussion centers on patterns of low-effort submissions that match common AI output styles.
Evidence of LLM Content in F-Droid
Commenters identified repeated app descriptions with identical sentence structures, generic feature lists, and placeholder icons. Several submissions used the same 200-300 word template across unrelated categories. F-Droid currently lists over 4,800 apps; thread participants flagged dozens of recent additions matching these markers.
Detection Methods Used by Reviewers
Community members described manual checks such as running text through open detectors and comparing code comments against typical LLM phrasing. One approach cross-references GitHub commit histories: apps with single-commit uploads and no prior activity raised immediate flags. No automated scanner is enforced at submission time.
Comparison With Other App Stores
| Store | Apps Listed | Reported AI Slop Cases | Review Process |
|---|---|---|---|
| F-Droid | 4,800+ | Dozens flagged | Volunteer, manual |
| Google Play | 3.5M+ | Thousands estimated | Automated + human |
| IzzyOnDroid | 1,200+ | Low | Curated fork of F-Droid |
F-Droid's open submission model lacks the scale of automated filters used by Google Play. IzzyOnDroid, a stricter fork, reports fewer low-quality entries due to tighter maintainer review.
Pros and Cons for Repository Quality
- Pros: Rapid growth in app count; some AI-assisted tools produce functional utilities.
- Cons: Increased maintenance load on volunteers; risk of broken or privacy-invasive code reaching users.
- Cons: Erosion of trust when users cannot distinguish hand-written from generated entries.
Who Should Pay Attention
Developers maintaining F-Droid forks or similar open repositories need updated review guidelines. End users seeking verified open-source apps should cross-check commit history before installation. Tool builders focused on code quality scanners gain a clear test case in F-Droid's current volume.
Practical Next Steps for Maintainers
Adopt lightweight checks: require at least three commits over 30 days and a human-written README exceeding 150 words. Integrate existing LLM detectors into the submission pipeline without blocking legitimate new contributors. Publish a public list of rejected patterns to deter repeat submissions.
Bottom line: F-Droid's volunteer model now faces measurable pressure from LLM-generated submissions that existing processes were not built to handle at scale.
The thread shows early consensus that stricter commit-history rules and basic text analysis can reduce slop without closing the repository to genuine contributors.
Top comments (0)