The Carpentries Generative AI Contributions Policy#
Last updated: 06 May 2026
This policy sets out expectations for contributions made to The Carpentries via generative AI tools, typically in the form of interactions with our public GitHub repositories.
Summary#
Contributors must acknowledge use of an LLM in the development of their contributions by adding “Assisted-by:
” (or similar wording) to their contribution. Contributions containing pasted output from an LLM, without effort on the part of the contributor to identify the most relevant parts and check their veracity, are not acceptable and will be hidden or closed.
Contributors must write issue text, comments, and pull request (PR) descriptions themselves; the assistance of LLM tools for translation or copy-editing is allowed.
AI-generated comments, discussions, and solutions to issues with the
good first issuelabel will be rejected.The use of agents to automate contributions is not allowed.
Maintainers are encouraged to deprioritise or reject contributions that they suspect have been made in violation of this policy.
Full Policy#
The context for this policy is set out in the Motivation section below.
Transparency#
Contributors must acknowledge use of an LLM in the development of their contributions by adding text similar to “Assisted-by:
Transparency about AI use helps the community understand the evolving role of these new tools and develop best practices around them.
Maintainers should label contributions that contain, or are believed to contain, generative AI tool outputs. Labelling will also facilitate reviews and set expectations for maintenance. Information about issue labelling can be found in our Maintainer Handbook.
Contributor Responsibility#
To ensure critical thinking, demonstrate understanding, and promote learning, contributors must write issue text, comments, and PR descriptions themselves (the assistance of LLM tools for translation or copy-editing is allowed). The contributor should be able to explain the motivation, implementation approach, expected impact, and any open questions or uncertainties within the issue, PR, or subsequent conversation with Maintainers or other contributors.
Review and Maintenance#
Contributors must read, review, and be able to explain all LLM-generated code or text before they ask other project members for review. The contributor is always the originating author and is fully responsible for their contributions. Contributors should be confident that the contribution is ready to be reviewed by a Maintainer with limited time, and they should be able to answer questions about their work during review.
New Contributors#
We expect that new contributors will be less confident in their contributions, and that is okay! We recommend that newcomers start with small contributions that they can fully understand to build confidence. For example, we mark issues with the “good first issue” label to highlight areas where people new to the community might be more comfortable contributing, and present opportunities for guided and collaborative co-working.
The Carpentries is dedicated to helping new contributors grow their expertise. As we teach in our workshops, learning is achieved through practice guided by formative feedback. AI tools cannot contribute to the learning opportunity presented by a “good first issue” and, therefore, AI-generated comments, discussions and solutions to issues with the “good first issue” label will be rejected.
Agents#
Agentic tools, such as the GitHub @claude agent, are designed to take action in digital spaces without human approval. Similarly, automated review tools are used to publish comments on issues and pull requests without human review. They represent a significant security risk and completely remove the human-in-the-loop following initial prompt setup. As such, contributions by agents within our Carpentries organisations and repositories are not allowed.
This excludes automated issues, PRs, and comments generated by Carpentries GitHub Actions and associated non-agentic workflows.
Handling Potential Violations#
If a Maintainer judges that a contribution does not comply with this policy, we advise that they use the following boilerplate response to request changes, further clarification, or human review:
This contribution does not appear to comply with our policy on generative AI content. It would require extensive review by us as Maintainers, and therefore requires additional justification before we would consider it for further discussion. Please read our developer policy on AI-generated contributions:
If a GitHub issue or PR is off-track, Maintainers should apply the “Extractive” label to help other reviewers prioritise their review time. They may also want to use the following boilerplate response:
We have assigned this contribution the “Extractive” tag. One of the best ways to make a change less extractive and more valuable is for you to perform a human review, reduce its size or complexity, or refocus it to directly address a specific community issue.
The factors that may make a contribution extractive are subjective and this policy leaves it up to the Maintainers of the project to ultimately decide. If a contributor fails to make their change meaningfully less extractive to the satisfaction of a Maintainer, we recommend that the PR or issue be locked for discussion and closed. We encourage Maintainers to discuss any issues or concerns they may have on our Slack channel or on a Maintainer call.
Copyright#
We aim to align this policy with our Privacy Policy and our transparent commitment to open source.
Artificial intelligence systems are testing our existing copyright laws, where corporate generative AI providers admit to breaking copyright to train models. The output of these models may well contain copyrighted material verbatim. As such, contributors are responsible for ensuring that they have the legal right to contribute code under the terms of our open source license. This typically means that either they, their employer, or their collaborators waive copyright of any intellectual property contributed to The Carpentries software and community.
Using AI tools to regenerate copyrighted material does not remove the copyright, and contributors are responsible for ensuring that such material does not appear in their contributions. Contributions found to violate copyright policy will be reported to our Carpentries Code of Conduct Committee and will be handled accordingly.
Accessibility and Inclusion#
The Carpentries champions Access for All and it is not our intention to exclude anybody from contributing to our open source projects through this policy.
AI assistance might take the form of feedback and suggestions to adjust wording and grammar, to draft alternative text descriptions, and to review contributions for dismissive language. Please note that we expect all contributions to be meaningfully reviewed by a human before they are submitted. Carpentries reviewers and Maintainers are often happy to provide this kind of guidance as well, if you choose not to make use of AI.
Motivation#
We believe that programming and lesson development are fundamentally creative processes. By using our collective theory of mind, humans provide a rich context for collaborating on and solving problems in society, including the logical and critical application of coding. We assume that we as The Carpentries community act in good faith, and always with a view to improve ourselves, our communities, and our society.
As such, our policy is underpinned by the following statements:
Contributions to our software projects need to be reviewed and maintained by a human. The use of generative or agentic AI to move the burden of software development to a maintainer is against our core values of empowering one another, putting people first, and championing community collaboration.
AI tools can be systematically or intrinsically biased against particular social, indigenous, and historically marginalised groups, due to the corporate ownership, geographic location, and ethical, legal, and health implications of their development and use. We advise contributors to weigh any benefits of using AI against the risks and harms it brings to those disenfranchised by proprietary and for-profit generative and agentic AI.
Demotivation, de-skilling and detrimental effects on mental health are proven by-products of the use of generative AI. This contributes to individual and wider social challenges, rather than easing them.
All contributions must involve a human-in-the-loop. Contributions containing pasted output from an LLM, without effort on the part of the contributor to extract the most relevant parts and check their veracity, are not acceptable and will be hidden or closed. We welcome open and transparent discussion to help ourselves and others learn, and the use of generative AI removes this ability to convert thought and experience into a formative discussion.
Processing contributions is not free - it takes Maintainer time and energy to review those contributions! Sending the unreviewed output of an LLM to open source project Maintainers extracts work from them in the form of design and code review, so this kind of contribution is called an “extractive contribution”.
Before the widespread use of LLMs, open source project maintainers would often review any and all changes sent to the project simply because posting a change for review was a sign of interest from a potential contributor. It was often a requirement that the contributor would remain active within the project to be on hand to continue to maintain the code they submitted.
While new generative tools enable development shortcuts, it shifts more of the effort from the implementor to the Maintainer. Our policy exists to ensure that we appropriately value Maintainer time.
The Carpentries encourages Maintainers to deprioritise or reject contributions that they suspect have been made in violation of this policy.
Reviewing changes from new contributors is a natural way to both sustain a project with new ideas and features, but also to find, foster and encourage the maintainers of the future. We want The Carpentries to be welcoming and open to aspiring open source contributors who are willing to invest time and effort to learn and grow, because growing our contributor base and recruiting new maintainers helps sustain the project over the long term. Being open to thoughtful contributions and helping people learn how to work with an open source project is a big part of how we have made an impact all over the world.
The Carpentries will always prioritise the humans that develop and maintain our open source projects and the humans that use and learn from them. This commitment to human collaboration will be reflected in the way that we document, curate and maintain our projects.
Acknowledgements#
This policy was heavily influenced by the LLVM AI Contribution Policy, and we would like to thank the authors for their clear and considered viewpoints.