Key details

  1. Flathub's current policy requires disclosure of known or reasonably suspected AI-generated code, documentation and packaging, with affected parts and approximate extent.
  2. AI-generated or AI-assisted Flatpak manifests remain prohibited even with disclosure.
  3. AI agents may not author or automate Flathub submission PRs, commit messages, descriptions, review comments or replies.
  4. AI-only research/debugging requires no disclosure if no generated material ships.
  5. Reviewers retain discretion to reject disclosed generated material; repeated misrepresentation can result in permanent submission bans.
  6. The policy revision was reported as amended September 4, 2026 and covered by independent Linux reporting on October 8.

What builders should take away

  1. Audit generated content before submitting; maintain a concrete disclosure record instead of relying on a blanket 'AI used' note.
  2. Treat Flatpak manifests and Flathub PR descriptions as human-written release artifacts.
  3. Do not let an AI agent open, write or review the Flathub submission PR.
  4. Expect reviewer discretion: disclosure makes a submission reviewable, not automatically acceptable.

What changed

Flathub's live app requirements replace the near-blanket generative-AI prohibition reported in May 2026 with a disclosure-and-review policy. Submitters must identify the AI-generated portions of code, documentation and packaging and their approximate extent. Reviewers may accept or reject that material at their discretion. Crucially, Flathub manifests must contain no AI-generated or AI-assisted content, and AI agents may not create or automate submission pull requests, write their commit messages or descriptions, or perform requested reviews. The revised text was reported as amended September 4 and received renewed specialist attention October 8.

Why it matters

Flathub is a significant distribution route for Flatpak desktop applications. For solo and open-source developers using coding assistants, the rule now distinguishes AI help in application implementation from AI help in release packaging and review. A project may be eligible despite some disclosed generated code, but a generated manifest or agent-authored PR can still block distribution. The policy also puts real documentation and human-review obligations on maintainers rather than promising automatic acceptance.

Disclosure is mandatory, not a safe harbour

The official requirements say submitters must disclose AI-generated code, documentation, packaging and other included material they know or reasonably believe was generated, identify the affected parts and give an approximate extent. Reviewers can reject a submission based on its generated material, maintainability, quality or the role AI played. An explicit disclosure does not create a presumption of acceptance. AI used only for research, discussion or debugging does not require disclosure when no generated material ships.

Manifests and submission pull requests remain human-authored

Flathub draws a stricter line around the repository's submission machinery. The manifest itself may not contain AI-generated or AI-assisted content even when disclosed. Agents may not open or automate submission pull requests, or generate commit messages, descriptions, review comments or replies; submitters may not request AI-agent reviews. That makes a fully automated AI coding-to-distribution pipeline incompatible with the published rules.

The change reverses May's broader prohibition

The earlier May policy was widely interpreted as banning AI-generated or AI-assisted application code apart from limited exceptions for mature projects. The revised September wording permits case-by-case consideration of disclosed material while keeping strict restrictions on manifests and submissions. Independent Linux specialists reported the reversal on October 8. This is a policy revision, not evidence that Flathub has begun accepting every AI-assisted app.

What developers should do before submitting

Document which files and changes used generated material, assess whether generated content is actually included, write and review the Flatpak manifest manually, and keep AI tools from authoring or automating the Flathub submission PR. Check the live requirements again before filing because reviewer discretion remains substantial. The general inclusion policy says it is not retroactively applied to older submissions, although later changes may trigger re-review.

What to watch next

  • Clarifications about how reviewers evaluate mixed human/AI codebases and generated documentation.
  • Examples of accepted or rejected submissions under the revised rules.
  • Whether Flathub changes its absolute restriction on AI-assisted manifests or PR authoring.

Still unclear

  • The rules leave 'approximate extent' and reviewer discretion deliberately open; no numerical threshold is published.
  • The live documentation does not provide acceptance-rate or enforcement data.
  • A developer's AI usage may be difficult to assess from repository history; the policy sets obligations rather than a technical detection mechanism.

Sources

Direct reading behind this dossier.

3 sources

Discussion

Discussion is reader-contributed. Comments are not part of the BTN dossier or its editorial evidence.

0 visible comments

Join the discussion

Keep comments useful and relevant. Reader contributions may be moderated and are not BTN editorial evidence.

Sign in to comment