ROS Project enforces strict rules for AI-assisted contributions
New guidelines require contributors to understand every line of code they submit.
The ROS Project Management Committee (PMC) has released official guidelines for AI-assisted contributions to ROS 2 repositories, extending the upstream OSRF (Open Source Robotics Foundation) policy. The TLDR is clear: feel free to use LLMs as development assistants, but when interacting on GitHub, Discourse, or Zulip, leave the AI-generated text behind. The guidelines emphasize that open source is about human collaboration, and maintainers want to engage with people, not chatbots. Maintainer time is finite, so submitting unverified LLM output shifts debugging and verification work onto them.
Key rules include the Author Ownership Rule: contributors must be the sole owners of their code and able to defend every technical choice. Zero tolerance exists for hallucinated APIs or configuration parameters; any PR introducing non-existent APIs will be closed immediately, with repeat offenses leading to blocks. Disclosure of AI use is mandatory for material contributions, and PR descriptions must be concise human-authored summaries, not generic AI phrasing like "enhances efficiency by optimizing asynchronous paradigms." Maintainers have full discretion to close low-effort AI dumps without review, ensuring the project remains human-centric and high-quality.
- Contributors must understand every line of code and cannot simply paste LLM output in response to maintainer feedback.
- Zero tolerance for hallucinated APIs or dependencies: PRs containing non-existent APIs are closed immediately; repeat offenses lead to bans.
- Disclosure of AI use is required in PR descriptions; generic AI-generated phrasing will be rejected in favor of concise human-authored summaries.
Why It Matters
Preserves open-source collaboration by ensuring human accountability and quality in AI-generated contributions to robotics.