PostmarketOS in 2026-02: generic kernels, bans use of generative AI

Original Article ↗ Hacker News Discussion ↗

Reactions to the AI Ban

  • Strong split: some celebrate an “uncompromising” stance with clear justification; others call it prejudiced, Luddite, or driven by culture war rather than pragmatism.
  • Supporters emphasize avoiding “slop” PRs and preserving reviewable, high-quality code, especially for kernel/drivers where mistakes are costly.
  • Critics argue the same complaints could apply to many past technologies and see it as fear of change rather than a rational tradeoff.

LLMs as Development Tools

  • Several kernel/low-level developers report using LLMs for exploration, boilerplate, and understanding unfamiliar subsystems, but not for “real” kernel-space or driver code.
  • Others say LLMs greatly boost productivity (2–5x more code), especially compared to waiting on forums like Stack Overflow.
  • Counterpoint: more code ≠ better software; cleaning up LLM output can be harder than writing from scratch, and reliance on LLMs may erode deep understanding.

Ethics, Ideology, and Licensing

  • The linked AI policy is described as primarily ethical (environment, labor, data exploitation), with code quality secondary.
  • Some argue open source has always been ideological; a project choosing not to use a tool is legitimate, and contributors self-select by values.
  • Others worry about unsettled licensing status of LLM-generated code.

Enforceability and Scope of the Ban

  • Policy bans both AI-generated contributions and recommending generative AI for postmarketOS problems.
  • Multiple commenters question how to distinguish AI-assisted code from “smart” autocomplete or normal reuse, calling enforcement effectively impossible.
  • Some say ignoring the policy while contributing would make you “a jerk”; if you dislike it, you should simply not participate.

Impact on Project Relevance and Velocity

  • One camp claims projects that avoid gen‑AI will become irrelevant as AI‑using competitors move faster.
  • The opposing view: postmarketOS targets niche, hard problems (e.g., mainline kernels, obscure device drivers) where LLMs are not yet decisive, and ethical choices can justify slower progress.
  • Debate on whether AI-free is meaningful given upstreams (Android, iOS, proprietary firmware) likely already contain AI-assisted code.

Device Support and Kernel Strategy

  • Brief technical side thread: comparison with LineageOS and AOSP kernels; postmarketOS can use newer mainline kernels on the same hardware because Android’s core features depend on eBPF and other AOSP patches.
  • Discussion that postmarketOS currently has few fully supported, recent devices; some question its overall relevance, others see it as valuable end-of-life support for older phones.

Broader OSS Maintenance Concerns

  • Several predict LLMs will flood maintainers with superficially OK but low-effort PRs, turning review into the main bottleneck.
  • Some foresee more projects either adopting similar bans or eventually closing to outside contributions altogether to cope.