Michigan Publishing Policies

Michigan Publishing Journals AI Statement
Last Updated: October 2026

1. MPS AI Policy

Michigan Publishing Services prioritizes the value of human skill, creativity, and critical thinking in the scholarly ecosystem. We recognize the expertise required in every step of the publishing process. Our policies seek to ensure the successful production and distribution of work within the scholarly community, and to maintain the integrity of the scholarly record. Currently, no AI tools are used at any point in MPS’s internal production management or publication workflows. Regarding the individual journals published by MPS, we adhere to the Committee on Publication Ethics’s position on artificial intelligence, while taking further cues from AI policies across the scholarly publishing ecosystem. Our policies are as follows:

Authorship

Across the spectrum, from commercial to noncommercial, publishers agree that AI cannot be listed as an author. Regardless of any contributions by an AI tool, the human authors remain responsible for the entire content of their published work and are liable for any breach of publication ethics, even if said breach was introduced by AI. 

Transparency & Disclosure

Authors must publicly disclose any use of AI in their work, including but not limited to: data analysis, image and/or text generation, abstract and keyword creation, language translation, bibliographic citation creation, copyediting, etc. Disclosure must indicate which AI tools were used, and where/how they impacted the work. 

Peer Review 

AI cannot be employed in lieu of human feedback during the peer review process. AI cannot peer review work and make a recommendation for the acceptance, revision, or rejection of an article. Many databases furthermore restrict or exclude journals that utilize AI in the peer review process, which hinders MPS’s ability to index our journals. On the grounds of confidentiality and copyright, reviewers may not upload authors’ works to AI tools. 

Production Services

In the interest of labor equity and quality control, fully AI-driven production workflows are not permitted in MPS journals. Copyediting must be performed by human professionals, not AI. If a journal wishes to incorporate AI tools during production, it must provide process documentation and sample content to MPS prior to implementing a new workflow. Should a journal proceed to utilize AI in their workflow (e.g., typesetting), this should be publicly disclosed on the journal’s webpage, either in the AI policy or in the submission guidelines. In addition, the journal must obtain consent from each author before uploading their work to any AI tools. Any AI-impacted elements must be subjected to the same level of human editorial scrutiny as any other published material–for example, if AI is used to translate an article into another language, the translated manuscript must be copyedited by a fluent reader of that language. 

Individual Journal Policy

Each journal must develop their own AI policy to specify if and how AI may be used in the creation or publication of a work. This AI policy must be publicly available on the journal’s webpage and referenced in the publication agreements signed between the journal and authors. Journal editors may find it helpful to reference the Library Publishing Coalition’s AI Editorial Policy Guide while developing their policy. 

2. Best Practices and Considerations

Because Michigan Publishing Services does not have editorial authority over our journal partners, it is each journal’s prerogative to decide what to allow or disallow in the content that they choose to publish. We encourage all of our journal partners to take the following topics and resources into account when developing their own AI policies:

Integrity

Large Language Models, or LLMs, create content based on patterns and probabilities. They do not truly understand either the data they’ve ingested or the results they output, and they won’t necessarily generate the same result when given the same input. As a result, GenAI content is subject to myriad issues, including but not limited to: factual inaccuracies (e.g., reproducing erroneous data or misinterpreting context), biases (e.g., reinforcing systemic prejudices or distortions in data), and confabulations or “hallucinations” (e.g., fabricating information to fill gaps in data). We endeavor to support the integrity of the scholarly ecosystem—from authors to publishers to librarians to students and scholars—by ensuring human skill and review is present in every stage of the process. There is broad agreement across the scholarly publishing field on this point: in addition to many noncommercial presses, commercial publishers like Elsevier and Wiley also ensure integrity by restricting usage and providing guidelines for their peer review process.

Labor

We acknowledge that many of our colleagues find LLMs to be useful tools for coding, accessibility, and data analysis. However, we are concerned about the disruption to ethical labor practices caused by the adoption of generative AI tools. For example, from global workers in the AI industry to the publishing professionals whose jobs are at stake (e.g., translators, graphic designers) to our colleagues at the Michigan Library system who are dealing with a deluge of AI content, artificial intelligence causes many negative effects on the value and dignity of human labor. 

Copyright

The question of copyright is subject to a fast-changing landscape. We stand on the side of author rights and encourage journals to stay abreast of rapid developments as they create their own policy and author agreements. LLMs are frequently trained on copyrighted materials, making it difficult to parse what is fair use and what may be considered plagiarism, especially when content is generated entirely via AI. Furthermore, if genAI is utilized in the journal’s workflow (e.g., typesetting), it is possible that LLMs may incorporate the analyzed text into their training data. For this reason, authors must consent to the use of AI in the workflow, per MPS policy.

Environment

The data centers powering GenAI pose a threat to our water systems and ecological health. U-M is committed to sustainability and carbon neutrality by 2040. We acknowledge the land stewardship of Indigenous nations, such as the Anishinabek (including Potawatomi, Chippewa/Ojibwe, and Odawa) tribal nations.

3. Recommendations for Editors

With the expectation that some journals may choose to embrace AI to one degree or another, we have drafted a series of recommended best practices in addition to our official policy. Regardless of the extent of AI use, human review is vital. If you are considering permitting AI use in published scholarship—or incorporating AI tools at any point in your pipeline—here are the pros and cons that we see for a number of potential use cases:

  1. Alt-Text Creation
    • Pros of AI use: 
      1. Allows quick and affordable compliance with accessibility requirements. 
      2. Particularly appealing for retroactively making archival content more accessible, which might be time- or cost-prohibitive to do manually (either due to the amount of work, or the difficulty of reaching authors years later).
    • Cons of AI use: 
      1. Authors know their work the best and can most accurately describe the content. 
      2. AI may misinterpret images and describe them incorrectly (e.g., hallucinations or biases). 
      3. Even if technically accurate, AI descriptions will not have context and may not focus on the relevant/important elements of the image as a result.
  2. Bibliography Creation/Formatting
    • Pros of AI use: 
      1. Theoretically makes it easier to follow the correct format for various citation styles.
    • Cons of AI use: 
      1. AI may fabricate new nonexistent references, or alter/combine/remove references that are already listed.
      2. Any errors in the bibliography make the rest of the paper appear suspect. 
      3. AI may mis-identify individual elements within each reference, so there’s no guarantee that citation styles are being followed correctly.
  3. Image Creation
    • Pros of AI use: 
      1. GenAI images (or text samples) may be necessary as illustrative figures in articles that are specifically about AI/LLM technology. Even if other AI use is disallowed, it’s worth carving out an exception for scholarly uses like this.
      2. A quick and affordable method for adding non-substantive, non-scholarly visual interest to a journal (e.g., article thumbnails, cover illustrations).
    • Cons of AI use: 
      1. May reflect poorly on the journal’s professionalism: many readers can visually identify AI images, and may perceive them as low-quality and low-effort (i.e., public backlash to “AI slop”). 
      2. Takes away jobs from artists and graphic designers.
  4. Keyword & Abstract Creation
    • Pros of AI use: 
      1. Probably the least-problematic potential use of AI in publishing, as keywords/abstracts are not per se part of the formal scholarship within the article. 
      2. AI may be less likely to introduce errors when distilling an article to a lower level of specificity (e.g., summarizing for an abstract or identifying frequently used keywords) as opposed to higher specificity (e.g., applying citation styles). 
      3. Takes some pressure off of authors.
    • Cons of AI use: 
      1. Authors know their work the best and can most accurately describe the content. 
      2. Confabulations and/or misinterpretations are still possible.
  5. Language Translation
    • Pros of AI use: 
      1. Makes article content more accessible to a wider variety of people by allowing it to be read in multiple languages. 
      2. Bypasses the language barrier for authors who may have otherwise faced obstacles to their work being published in their desired journals.
    • Cons of AI use: 
      1. Takes jobs away from skilled translators. 
      2. Inaccuracies may be introduced during translation, due to either AI fabrications/biases or “lost in translation” misunderstandings. 
      3. Any AI translation must undergo copyedit reviews by a human fluent in the destination language. 
      4. May not be worth the effort, as readers can just do their own machine translation if they want (e.g., Google Translate). 

4. Additional Resources

Michigan Publishing Journals Preservation Statement
Last Updated: July 2023

1. Introduction

This statement outlines the preservation policy of the Journals Program of Michigan Publishing. This policy covers serials both managed within Michigan Publishing and published on behalf of our publishing partners. It is adapted from the NASIG Model Preservation Policy and conforms to the Digital Repository Services Digital Preservation Policy of University of Michigan Library, of which Michigan Publishing is a part. 

This policy articulates a commitment to making the work that we publish today part of a durable scholarly record that benefits the authors, readers, and scholars of tomorrow.

2. Scope

2.1 What we preserve

The primary focus of our digital preservation activities is preserving the final published version of the content we publish. The final published version is the edited and formatted version of the article, including any digital resources or supplemental material in one of our managed hosting environments on Janeway or Fulcrium. 

Articles (in JATS and PDF format) and accompanying images are published on Janeway, an open source scholarly publishing platform hosted and supported by The Open Library for the Humanities.

Any audio, video, or interactive map resource files are hosted on Fulcrum, managed and hosted by Michigan Publishing. The JATS and PDFs on Janeway will link to these resources and, when appropriate, may include embed codes to allow for their inline display.

2.2 What we do not preserve

Our preservation commitments do not at present attempt to capture the context or user experience in which the materials described above are made available to the public, nor any materials produced in the course of creating the final published record. The following are considered out of scope:

  • Manuscripts submitted to our system that have not been accepted for publication
  • Versions prior to the final published version of accepted manuscripts
  • Peer review notes and author responses
  • Comments
  • Related blog/social media posts or other promotional material
  • Content linked to but not hosted on Janeway or Fulcrum

Any annotations made using hypothes.is are not currently covered by this commitment, and hypothes.is does make its own preservation commitment; see their Terms of Service. Expanding our preservation commitment to include these annotations is a subject of interest and future investigation.

3. Strategies

3.1 Storage and Integrity

Files hosted on Janeway are managed in the Open Library for the Humanities storage and backup environment. All JATS XML, PDFs, and images are also deposited into CLOCKSS.

Materials hosted on Fulcrum are managed according to the Fulcrum Preservation Policy.

Through our hosting partners, Michigan Publishing is employing a number of methods to achieve bit and migration preservation strategies. For details, please refer to their specific preservation statements linked above.

 3.2 Access

Michigan Publishing commits to maintaining access to materials as part of the durable record of scholarship. This includes migration of content to new hosting environments as technology and vendor partnerships require. All articles and resources are assigned DOIs and metadata deposited at Crossref, and DOIs will be updated in accordance with any changes in hosting environments, in order to ensure citations remain viable in the future.

Many of our journals are published by arrangement with editorial teams that are outside our organization. In the event that these partners choose to end their hosting agreement with Michigan Publishing, we will provide all preserved content to assist with migration to a successor hosting partner, with the partner assuming responsibility for preservation past that point. If a partner suspends or ceases publication of a journal, Michigan Publishing will evaluate its options for either providing ongoing access to this content or effecting a triggering event to provide access through CLOCKSS.

3.3 Management

This policy is maintained and managed by the Director of Publishing Services and Director of Publishing Technology, under the direction of the Associate University Librarian for Publishing / Director, University of Michigan Press, University of Michigan Library. It will be formally reviewed every 3 years, and may be updated to reflect changes of an organizational, technological, or environmental nature. All active partners will be notified of any substantial updates to this policy. All changes are made in consultation with the Digital Repository Services Steering Committee and Digital Preservation Steering Committee of University of Michigan Library.

4. Principles

Michigan Publishing’s digital preservation policy is guided by the following principles:

  • Our digital preservation strategies are directed by our mandate to build, steward, and preserve our collections on behalf of our user community. 
  • Our digital preservation arrangements will comply with our existing policies, procedures, and/or relevant applicable laws and regulations.
  • We affirm that the purpose of digital preservation is to maintain long-term access to digital assets over time. 
  • Digital preservation is a resource-constrained activity. Choices to implement arrangements or infrastructure must be made in the context of our overall resourcing environment and seek technologically, organizationally, and financially sustainable solutions, including through collaboration and cooperation within the organization and with external organizations and collaborative networks. 
  • We will prioritize the use of platforms and tools from which preserved objects and all associated metadata can be extracted with integrity.
  • We will demonstrate leadership and support development in the digital preservation field within its financial and staff resources by:
    • Using open-source software;
    • Contributing to standards creation;
    • Participating in scholarly and practical discussion;
    • Using persistent identifiers such as DOIs, ISSNs, ORCiDs etc.