[ADD] AI POLICY

[etobella]

OCA has experienced contribution patterns that are not sustainable, such as high volumes of reviews in very short periods - in one case 25 pull request reviews in a single hour, and elsewhere a month's worth of review activity arriving within a couple of days. As a consequence, other community members spend time and energy responding, and the team does not have the capacity to manage this level of work at the standard of quality and professionalism the OCA requires.

We recognise that AI can be of great advantage in producing code. The interaction with the ecosystem - review, discussion, judgement, accountability - must be human.

The legal questions around AI and copyright remain unresolved. Whether AI can author a work, and if not, whether authorship falls to the contributor, the AI maker, or the creators of the material the AI drew from, is not settled in law. OCA takes no position on authorship however OCA's position on responsibility is absolute - the human contributor is responsible for their contribution in full, however it was produced. These are two distinct roles - authorship may one day be defined by others; responsibility is defined in this policy.

AI tools can generate and review code yet what they cannot do is understand the culture of this community, the interdependencies of the ecosystem, or the impact of their output on the volunteers who must validate and commit it. That understanding is human work.

We are not judging the tools you use to create and review code, but we are very clear on what we will accept, both technically and in the way you engage with the community.

It is never the AI's fault, as AI is not intelligent. A human instructs, connects, installs and directs it. When a human disrespects the ecosystem through their contributions and use of these tools, that human will be held to account.

This policy is effective immediately.

TL;DR

• We know that you might use AI tools to help you.
• We don't accept abdication of responsibility to AI.
• Every contribution must come from a human who understands it and takes full responsibility, however it was produced.
• Any AI involvement in authoring means the contribution should include an Assisted-by: trailer - it's binary, it either had AI involvement or it didn't, with no threshold to judge.
• AI tools never go in Co-authored-by:.
• Unsupervised agentic tools are not permitted.
• If you can't explain and defend every line, don't submit it.
• During review: engage with feedback, don't just regenerate and resubmit. "The AI wrote it" isn't a valid answer.
• When reviewing, don't post AI-generated review comments or summaries without fact-checking them yourself.
• Consequences: rejection, permanent ban for bot-like behaviour, or being blocked until you talk to the Governance Working Group.

@sbidoul @StuartJMackintosh @OSevangelist

[rousseldenis]

Many thanks to those who established this!