rfc: add organization-wide AI policy RFC#31
Conversation
This RFC proposes an org-wide policy governing the use of AI coding assistants across all ODP repositories, covering human accountability, legal compliance, review/testing standards, enforcement, and integration with contribution guidelines. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
jerrysxie
assigned
jerrysxie and unassigned
dhillonk,
spbrogan,
maxperga-msft and
jerrysxie
|
Please review and provide any feedback @OpenDevicePartnership/odp-platform-maintainers |
| 4. **No direct AI submissions**: ODP does not accept contributions submitted directly by AI tools without | ||
| substantive human involvement. Automated pull requests generated entirely by AI agents without human review are | ||
| not permitted. |
There was a problem hiding this comment.
I would prefer that we do not allow automated pull requests generated by AI agents at all regardless of human review. Many of the other requirements are more difficult to enforce and trace in history without a human using the AI only as an assistant and fully attesting to its output.
I believe the "Proposed Policy" section later follows this approach anyway.
There was a problem hiding this comment.
It is very likely that we will have automated pull requests generated by AI to update documentation, workflows, and other accompanying files. Do we want to restrict the stronger language for source code contribution only?
There was a problem hiding this comment.
In Patina, we distinguish between automated bots and AI. We use automation with bot accounts to perform those tasks, but we still do not want to have an AI driven bot directly author PRs.
Are you still referring strictly to AI in this context vs automation? If so, this might be an area that differs for Patina.
felipebalbi
left a comment
felipebalbi
left a comment
There was a problem hiding this comment.
One comment below, all in all, looks good.
There was a problem hiding this comment.
Pull request overview
Adds a draft RFC proposing an organization-wide policy for using AI coding assistants across Open Device Partnership (ODP) repositories, focusing on contributor accountability, legal compliance, and quality expectations.
Changes:
- Introduces a new RFC document defining goals, requirements, prior art, and a proposed org-wide AI policy.
- Outlines enforcement expectations and how projects can extend the baseline policy.
- Proposes integration options for surfacing the policy via per-repo
CONTRIBUTING.mdand/or a standalone governance document.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
- Add 'As of today' qualifier to AI tool categories (review OpenDevicePartnership#1) - Add device testing requirements where applicable to Policy §3 (review #3, OpenDevicePartnership#10) - Resolve unresolved question on policy tiers: single baseline + extensions (review #5, OpenDevicePartnership#8) - Resolve unresolved question on PR checkbox: not needed (review OpenDevicePartnership#9)
* Added Linux Foundation generative AI guidance section to provide additional context on how other open source organizations handle AI-generated contributions. * Standardized the `Assisted-by:` commit trailer guidance to align with the Linux kernel approach,
This RFC proposes an org-wide policy governing the use of AI coding assistants across all ODP repositories, covering human accountability, legal compliance, review/testing standards, enforcement, and integration with contribution guidelines.