Skip to content

AI Contribution Overview

When to Use

Read this first. It maps the landscape of AI contribution policies, why disclosure matters, and how Drupal's "disclose and verify" position works before you touch any issue or merge request.

Decision

Project Position Policy
QEMU Ban AI-generated code prohibited entirely (June 2025, DCO rationale)
Gentoo Ban AI-generated code banned (April 2024, 6-0 council vote)
Linux kernel Disclose Co-developed-by trailer required, same review standards apply
Apache Disclose Generated-by convention, AI welcome with attribution
Drupal Disclose and verify AI welcome with mandatory disclosure, human responsibility

Pattern

Drupal's four disclosure levels:

Level Definition
AI Assisted Issue AI helped research or write the issue
AI Assisted Code Human wrote code with AI suggestions, reviewed every line
AI Generated Code AI generated substantial code, human reviewed and tested
Vibe Coded AI generated most/all code with minimal human review

The responsible path: AI Assisted Code with thorough human review.

Core principles: 1. Disclosure is mandatory — every issue and MR must indicate AI involvement 2. The human is responsible — contributor takes full responsibility regardless of how code was produced 3. Review expectations scale with AI involvement — Vibe Coded triggers maximum scrutiny 4. Policy is evolving — check drupal.org for the latest governance decisions

Common Mistakes

  • Wrong: Not disclosing at all → Right: Even minor AI usage (autocomplete) should be disclosed; when in doubt, disclose
  • Wrong: Treating disclosure as optional → Right: Undisclosed AI usage erodes trust and may lead to credit revocation
  • Wrong: Assuming "Vibe Coded" is acceptable → Right: Contributors who cannot explain their own code face maximum scrutiny and likely rejection
  • Wrong: Confusing Drupal's openness with permissiveness → Right: "AI welcome" does not mean "anything goes" — quality standards are the same or higher

See Also