Hi everyone,
We have something new to share: PKP has published a policy on the use of generative AI in contributions to our software.
You can read it here: https://pkp.sfu.ca/pkp-policy-ai/
Why we wrote it
Generative AI tools have become part of how many people work. That’s not something we wanted to ignore or leave ambiguous. We’d rather be upfront about what we welcome, what we ask of contributors, and why those expectations matter for the health of our shared project.
The policy comes from a recognition that open source software depends on careful, accountable contribution, and that AI tools, used thoughtlessly, can create real burdens for the people doing maintenance and review work.
What it covers
A few things worth highlighting:
Disclosure is required. If you use generative AI in a contribution, whether that’s code, documentation, or a bug report, please tell us. A brief note explaining what you used and how goes a long way toward helping us review it appropriately.
-
A qualified human in the loop is always required. AI-generated contributions should be reviewed, tested, and understood by the person submitting them. We shouldn’t be receiving code that the contributor can’t speak to or test results that haven’t been verified.
-
Extractive contributions. This is a concept we found useful and wanted to name explicitly. An extractive contribution is one where the cost of reviewing and integrating it outweighs its benefit to the project. This isn’t unique to AI; any contribution can be extractive, but generative AI makes it easy to produce large volumes of confident-looking output quickly, which increases the risk. We ask that all contributors, AI-assisted or not, be mindful of this.
-
Contributions as learning spaces. PKP puts real effort into supporting new contributors. Using AI to shortcut that process doesn’t just affect the quality of the contribution; it affects the contributor’s own learning, and the sustainability of our contributor community over time.
What this doesn’t mean
This policy is not a ban on using AI for software contributions. It’s an invitation to use them responsibly and in ways that keep our community strong.
We’re genuinely interested in your thoughts on this, particularly from those of you who contribute code or documentation regularly.