Best Practices for AI and Automated Tools in Open Access Publishing: DOAJ Guidelines for Authors
TL;DR
- AI use is allowed in open access publishing with proper adherence to DOAJ guidelines as well as individual journal’s policy.
- AI disclosure statement, verification of AI output, no authorship credits to an AI tool, and not listing a generative AI tool as a source are the four key requirements for publishing in OA journals.
- It is best to restrict AI use to brainstorming, outlining, structuring, and improving rather than using it to “generate” research ideas or entire manuscripts.
Open access publishing has evolved over the years, revolutionizing the way research is shared and accessed globally. Open access (OA) journals create opportunities for collaborations and innovations by allowing readers access to scientific knowledge without restricting it behind paywalls, encouraging researchers to further cross-disciplinary science.
But alongside scientific publishing, we’ve seen a dramatic evolution of Artificial Intelligence (AI) and automated tools, which are now well and truly a part of the research landscape. With increasing concerns around AI-generated papers, could authors exploit AI tools to get publications out there? Not if the Directory of Open Access Journals (DOAJ) can help it!
This article looks at the AI use guidelines set forth by DOAJ (for journals and for authors) and outlines some best practices for ethical and responsible use of AI tools in OA publishing.
Open Access AI Guidelines for Authors
Open Access AI Guidelines for Journals
Best Practices for AI Use in OA Publishing
FAQs about AI Use in OA Publishing
What is OA Publishing?
Publishing in OA journals simply means your research can be read by users for free. They do not have to pay any subscription fee to access the papers published in OA journals, which means your research is likely to receive more eyeballs and perhaps even citations owing to its reach and availability.
Publishing in open access journals comes with benefits and risks alike. Authors can reach a much wider audience than when publishing in a subscription-based journal. The time to publication is also comparatively shorter. So time-sensitive research that report crucial findings can be made publicly available at a much faster rate. Yet, authors voice concerns around the quality of journal:
“What if my paper ends up in a poor-quality publication?”
“How do I know this is not a predatory journal?”
We’ve previously addressed these concerns and you can read all about how to find the correct open access journals, how to avoid predatory publishers, and what differs between subscription-based and OA journals.
Here, we will look at the newly evolved expectations in OA publishing: The AI use policy. Regardless of the subscription type, most major publishers now outline guidelines for ethical use of AI in journal publishing considering the integration of AI tools into the research workflow. DOAJ divides the guideline into two categories: one for authors submitting to OA journals and the other for the journals themselves.
Open Access AI Guidelines for Authors
Like subscription-based journals, OA journals must formulate their own AI use guidelines for authors before registering under DOAJ. Here are a few key pointers to note.
- You must properly disclose AI use. OA journals require authors to use AI disclosure statements during submission. Specify the tool name with version, why it was used, and whether the AI output was validated. It is not only for language and grammar editing. You could have used an AI tool like R Discovery during literature review or a combination of automated tools for creating figures. These must be disclosed too. However, the use of inbuilt checkers like Editor on Microsoft Word need not be disclosed.
- You must verify all AI output. Using AI tools for assistance is not wrong; but blindly trusting the AI-generated output and not verifying its accuracy is careless.
Picture this: generative AI tools like ChatGPT or Claude takes your figure, and in the process of enhancing one part ends up modifying the labels in the figure into gibberish text. Now, as authors, if you are satisfied with whatever enhancement you requested and fail to check the remaining portions of the figure and submit it as is to the OA journal, you’ve basically shown that you did not completely verify the AI output. This is unprofessional and could have serious ramifications if the paper gets published. - Do not give authorship credits to an AI tool. Say a tool like Paperpal was used to prepare your manuscript. Can you credit it as an author? Absolutely not! Authorship must always remain with humans as per the Committee on Publication Ethics (COPE) guidelines [1]. No OA journal will accept any AI tool to be one of the co-authors of a research paper. So, even if you’ve utilized an AI writing tool, simply disclose its usage and how much it contributed to manuscript preparation but do not mention it as a co-author.
- Generative AI is not considered a valid source. When listing references and citing your sources, avoid stating “ChatGPT” or “Claude” as your source! If you’ve received information from such gen AI tools, then you must dig into the references the tool has cited and verify that they are indeed real and not fabricated by the tool. The cited source must be the original work and not the name of the tool that led you to that work.
Open Access AI Guidelines for Journals
One of the first requirements for journals to be indexed in DOAJ is for them to have their own AI use policy for authors. And these should entail the mandatory AI guidelines for authors stated above. In addition, journals can have further requirements, such as allowing AI use only for manuscript writing and editing but not for figure preparation.
The guidelines also extend to journal editors and peer reviewers. DOAJ does not allow the use of generative AI for peer reviewing or manuscript assessing by editors because there’s a risk of confidentiality being breached. Feedback and evaluation from generative AI could also be biased depending on the training datasets used for the large language models (LLMs). Some OA journals may allow the use of generative AI tools for only compiling and editing the peer review report with proper disclosures. And another key thing to note is DOAJ strongly encourages the human-in-the-loop model, wherein humans closely monitor any automated task or AI-powered workflow. This means that with solid verification and validation in place, OA journals can integrate AI tools and automation into their publication workflow.
What does this mean for authors? Well, do your homework on the preferred OA journal. What’s allowed? What’s not? Does the journal align its AI policy with that of DOAJ? All these must be checked before finalizing the journal.
Best Practices for AI Use in OA Publishing
DOAJ outlines specific guidelines for authors, editors, peer reviewers, and journals applying for OA status [2]. What’s key for authors, however, is to follow certain best practices when planning to publish in OA journals.
1. Know your journal’s policy. First, confirm whether the OA journal you are targeting adheres to the DOAJ’s AI use policy. Then, check the journal’s author guidelines. These should align with the guidelines recommended by DOAJ. This helps ensure that you are following proper requirements when preparing your manuscript.
2. Limit use of AI tools for brainstorming and structuring. Not all tasks are meant to be automated. Rather than allowing AI influencing your work, retain your authority when steering idea generation, interpreting findings, and analyzing study implications. AI assistance can be restricted to checking your writing and fixing grammar errors. But the tone of writing? Keep it your own!
3. Always validate AI-generated output. Do not accept AI output at face value. Generative AI tools could be inaccurate, fabricate citations, or hallucinate information to fit the output to the prompt input by the user. So, double check that the AI claims are indeed genuine.
4. Format manuscripts with minimal AI assistance. When you are formatting manuscript for OA journals, pay attention to journal-specific guidelines rather than instructing a generic AI tool to give you a formatted manuscript. Basic layout requirements like margin setting, page setting, page/line numbering can perhaps be verified by AI tools. But checks involving authorship, wording of disclosure statements, placement of acknowledgments, word/character limit of title/headings/sub-headings, requirement of running title, abstract (structured or unstructured)—all warrant careful human intervention.
5. Disclose AI use transparently. If you’ve used AI anywhere in your research workflow, honestly reveal it to OA journals during submission. There’s no point in hiding. Responsible disclosure is a part of ethical AI use. So ensure that you provide the necessary details of how AI assisted in your manuscript preparation.
6. Do not make AI the co-author. Listing an AI tool as an author is not allowed because a tool cannot take ownership of what is conveyed in a research paper. The ultimate responsibility lies with human authors and not AI writing tools. So, even if there was extensive use of AI during manuscript writing and refining, avoid naming it as the author.
With journals normalizing the use of AI and automated tools, authors must know what’s acceptable and what’s not. If you are unsure, get Editage’s Post-AI Risk Assessment service for your AI-assisted manuscript and check what AI missed!
FAQs about AI Use in OA Publishing
1. Does DOAJ allow the use of AI tools?
Journals have to implement the AI use policy approved by DOAJ, which includes a few mandatory requirements. Authors can use AI tools when submitting research papers provided they disclose its use appropriately, verify all AI-generated output, do not list an AI tool as a co-author, and avoid mentioning generative AI as a valid source in their manuscript.
2. Will open access journals accept AI-authored papers?
No. As per DOAJ AI guidelines, AI cannot be considered as an author or a co-author of a research paper. AI can assist in drafting manuscripts but should not be given authorshipp credits.
3. Is there a risk to using AI tools when writing research papers?
The level of risk involved in using AI tools depends on how they are used and to what extent. If you over-rely on generative AI tools like ChatGPT, there’s a risk of citation fabrication, data falsification, plagiarism, or AI hallucination. So, authors must be aware of how much an AI tool is contributing to their research. AI-generated information should not overshadow authors’ own writing.
4. DO open access journals use AI tools for peer review?
As per DOAJ guidelines, open access journals must avoid using AI-powered peer reviews since maintaining confidentiality of unpublishing scientific research is key. However, some journals may allow the use of AI tools for compiling the peer review reports after the reviewers have evaluated the submitted manuscripts.
References
1. Authorship and AI tools: COPE https://publicationethics.org/guidance/cope-position/authorship-and-ai-tools
2. DOAJ Guide to Applying https://blog.doaj.org/2025/11/03/updates-to-the-guide-to-applying-flipped-journals-ai-quality-criteria-and-oa-exceptions/
