Introduction
Since 2022 when ChatGPT, Claude, Copilot, Gemini and other tools powered by large language models (also known as generative artificial intelligence, GenAI, or simply AI) started to be made publicly available, OER Commons staff have seen a sharp rise in the use of these tools to create open educational resources. We believe that adapting new tools and technologies to support educators and learners, when done with care and intention, can lead to better teaching and learning experiences.
However, we’re concerned about the ways that unchecked AI use can lead to an overall decrease in the quality of resources across the OER landscape. As librarians of a digital public library serving tens of thousands visitors a day, we have a responsibility to our community to steward a healthy OER landscape. In this spirit, we’ve put together this guidance so that authors and the wider community understand our expectations regarding AI use in the materials submitted for inclusion on OER Commons.
These expectations have been developed to align with OER Commons’ Collection Policy, as well as with ISKME’s Guiding Principles for Responsible AI in the World of Open. They aren’t meant to replace or supersede the policy so much as to help users understand how we’re interpreting and applying the existing policy when it comes to AI and OER creation.
In short: Effective August 1 2026, we expect all users to:
- Disclose or attribute AI when it has been used in the creation of the resource they’re submitting;
- Avoid submitting tools or websites created through wholesale use of AI that lack meaningful and verifiable human attribution; and,
- Review and edit any AI-created content thoroughly before submitting it for consideration.
Failing to take the above steps will likely result in a user’s submitted resources being rejected. Because of the high volume of resources our librarians are tasked with reviewing every day, we can’t provide specific feedback every time a resource is rejected by our team, so we encourage everyone to review this document. Section 3 “Materials Selection” of our collection policy goes into greater detail on how we make decisions which resources to include in the collection or not.
Read on to understand our rationale behind the above expectations. A selection of related readings is available at the end of the document as well.
AI Use Disclosure: No AI tools were used in the authoring of this document.1. Disclose any AI use
If you used AI tools (e.g. to generate text, an outline, images, or other content) in the creation of your resource, this should be disclosed in your resource’s description when you submit it to OER Commons. This serves several purposes.
First, disclosure makes it easier for our staff to identify AI co-created resources and curate them appropriately (e.g. into the AI & OER Hub). To reiterate: Using AI in the resource creation process is not itself a reason that we would reject a resource from being included in our collections; we will continue to use our collection policy as the guiding document when making collection decisions.
Second, it provides transparency about your resource to users who may have concerns about (or policies against) using things created with AI in their classrooms. Whether or not one subscribes to the basis of those concerns, we want to ensure that the OER ecosystem we foster is based on mutual trust and respect. We respect that authors may be concerned that identifying their work as created with AI will result in their materials not being adopted, but this concern does not override a person’s right to informed consent.
Third, the current state of U.S. copyright law is that AI-generated materials are not copyrightable since only human-generated works can be copyrighted. This may impact the applicability of the license you select.
There are many ways to appropriately disclose AI use in the creation of your resource. AI Use Disclosure resources is a lesson that provides information about several ways to properly attribute AI use (however, please note that adding the tag “AI Co-Authored” as shown in the resource is welcome, but on its own not enough to satisfy disclosure requirements). Proper disclosure can be as simple as a couple of sentences explaining how you used AI to create your item.
Disclosure examples
Our team is currently working on making it easier to self-disclose AI use when you submit a resource; for now, here are some example statements that you can take, modify/remix, and add to your resource’s description.
AI used to draft text
AI use statement ChatGPT was used to create the first draft of this resource’s text. Minor edits to machine-provided text were made by the author. Images found in this resource are not machine-generated, to the best of the author’s knowledge.
AI used to create images/draft outline
AI use statement Gemini (Nano Banana) was used to create images found in this resource, as well as in drafting the outline of this resource. The majority of the text was written by the author, with minor edits for clarity and grammar provided by Gemini. Source materials include [names of/links to source materials].
AI used to create a tool/website
AI use statement This tool was created in Claude Code using Claude Opus 4.8 with guidance from the author. The code base of the resource can be found at github.com/username/example-guthub-repository-link for remixing.
AI used to aggregate info from different documents
AI use statement NotebookLM was used to create this resource by providing it the following source materials: [insert names/urls here, or make reference to the section of the resource that provides citations]. Minor edits to machine-provided text were made by the author.
AI used to translate between languages
AI use statement Portions of this resource were translated with the assistance of ChatGPT to [target language] from [source language]. Minor edits to machine-provided text were made by the author.
Our staff review all submitted items. If you submit an item to OER Commons that we have reason to believe was created at least in part with AI tools but you do not add a disclosure to the description, our staff reserve the right to reject your submission.
2. Wholesale AI-generated tools and websites will be rejected
AI tools have enabled people who previously lacked the time or technical expertise to create fully (or mostly) functioning interactive websites, online tools, large quantities of content, and more in short amounts of time. These tools and content may have legitimate educational uses, but if we believe (based on appearance, functionality, quality, or other indicators) that these tools and websites have been primarily created by an AI tool with minimal involvement from a human beyond the prompting used to generate the content, we will reject these resources.
OER Commons exists as a public library to benefit teachers and learners by making quality OER more easily discoverable. It has done so since 2007, and as a result other sites linked from our library may benefit from improved search presence. However, adding a resource on your own site to OER Commons’ collection because you want to improve your site’s searchability when you are otherwise uninvolved in education is a misuse of our credibility. When we review a website that has very little to no information suggesting that it was created by someone with expertise or history in education, little to no verifiable information about its author, or little to no information about how the content itself was created or reviewed, this is a red flag to us. We’ll likely assume that the submitter is trying to game online search and discovery, reject the user’s submitted resource, and possibly remove their account.
We acknowledge that one of the most valuable things about OER is that it lets people with diverse backgrounds and skills use them to create, share, and remix materials. Our staff have to be able to verify an item’s quality, and part of that includes understanding who created it and how. In line with that, if we review a resource that has hundreds of pages of content but find that its authorship is limited to one or two people (or has no clear and verifiable human attribution), or created in a short amount of time, this prevents us from making a determination about the resource’s quality, authorship, or other criteria outlined in our collection policy. Our staff will likely reject it.
If you’re new to creating OER, we welcome you to explore our collections, see how others have used Open Author to create resources, or try your hand at creating OER with Open Author yourself.
3. Thoroughly review AI-created content before submitting
All educational resources – and especially open educational resources – serve learners best when they are written or designed with real learners in mind. When AI is involved in creating OER, the role of a human in that design process becomes even more important.
AI tools can produce large quantities of plausible text, and since 2022 the factuality of responses has markedly improved across the board. However, high quality educational resources are more than just factually accurate: they are created explicitly to help convey knowledge and information. This can be done in a number of ways: by encouraging self-reflection, including interactive elements, writing content at age-appropriate levels, building in assessments, etc… It can also be done by creating resources that reflect the lived experiences of those who create them, as well as those who will learn from them.
Anyone can ask an AI tool to write a lesson plan geared toward a given kind of learner for a given topic, and the tool may return a response that on its surface appears to match the criteria. If the asker is an experienced educator, this could be a valuable first step in the creation of something well-designed for learners. This is a great use case because it permits someone with expertise to bring that to bear while also lightening their load. However, AI tools on their own bring no personal lived experience to the authoring process. When used to create large quantities of content, this makes it less likely that the creator will be likely to review all of the tool’s outputs (or may not be able to, depending on the quantity, the form of the generated content, and the prompter’s own expertise).
This increases the likelihood that the text contains factual errors, incomplete details, and other issues that would otherwise be caught in the process of editing. Similarly, excessive use of AI tools to create bodies of text serve to uplift the voices and perspectives contained in the texts that the models were trained on, de-emphasizing more marginalized points of view and homogenizing language use generally, resulting in the existence of a sort of “AI voice.”
A similar “AI voice” is also present when AI tools are used to create online tools or websites. The generated site may appear “good enough” at first glance, but in-depth use or exploration of the site will reveal unnecessary or poor design choices, incomplete pages, nonfunctioning links, contradictory terms of use / “legalese” with little basis in reality, etc. These phenomena have been documented extensively, and warrant meaningful consideration by anyone creating educational resources using AI. We encourage you to review the additional readings at the end of this document for more information
Because of these and other concerns around quality, resources that appear to have not been edited in a meaningful way after being generated from an AI tool are likely to be rejected.
We believe that AI can be a boon to the creation of high quality OER, but that depends on authors being judicious and transparent about its use, and realistic about its limits and impacts on the learning experience. Our expectations may evolve as these technologies change, but we hope that our own transparency in this lets OER creators feel more comfortable when creating and submitting their works to our collection.
For resources on OER Commons that have already been identified as having been created with AI, we'll be reaching out to users directly and asking them to update their submissions with appropriate attribution over the next few months.
Additional readings
OER Commons collection policy. Institute for the Study of Knowledge Management in Education, 11/21/2024
ISKME’s Guiding Principles for Responsible AI in the World of Open. Institute for the Study of Knowledge Management in Education, 11/21/2024
The creative link between words and ideas is weakening in the AI Era. Moon, K., Kushlev, K., Bank, A., Luttges, B. L., Viskontas, I., Kaufman, J. C., … Green, A. 2/5/2026
Generative AI enhances individual creativity but reduces the collective diversity of novel content. Anil R. Doshi, Oliver P. Hauser. Science Advances, Vol 10, Issue 28. 12 Jul 2024
A New Direction for students in an AI world: Prosper, prepare, protect. Mary Burns, Rebecca Winthrop, Natasha Luther, Emma Venetis, and Rida Karim. (Brookings Institute). 1/14/2026
Teachers save time with AI. Their students may pay the price. Jill Barshay. (The Hechinger Report) 7/13/2026