Use of artificial intelligence tools in the publishing process: expectations from publishers through author guidelines

May combine two claims.

Taylor and Francis highlighted the opportunities of AI in idea generation and overcoming language barriers, a more permissive stance compared to Emerald, which strictly prohibited AI-generated abstracts and literature reviews.

“For example, Taylor and Francis highlighted the opportunities of AI in idea generation and overcoming language barriers, a more permissive stance compared to Emerald, which strictly prohibited AI-generated abstracts and literature reviews, as shown in Table 2.”

May combine two claims.

Sage distinguishes between assistive AI tools such as Grammarly and language improvement software, and generative AI tools such as ChatGPT or DALL-E, noting that only the latter category requires mandatory disclosure.

“Sage further distinguishes between assistive AI tools such as Grammarly and language improvement software, and generative AI tools such as ChatGPT or DALL-E, noting that only the latter category requires mandatory disclosure.”

May combine two claims.

Emerald permits the submission and publication of AI-created images but requires that they be flagged and cited.

“Emerald permits the submission and publication of AI-created images but requires that they be flagged and cited.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Frontiers supports responsible AI use that includes language improvement provided confidentiality and copyright protections are respected.

“Frontiers supports responsible use that includes language improvement provided confidentiality and copyright protections are respected”

May combine two claims.

Sage states that authors must indicate the use of AI in the manuscript and clearly reveal AI-generated content along with the submission.

“Sage states that authors must indicate the use of AI in the manuscript and clearly reveal AI-generated content along with the submission.”

May combine two claims.

COPE's position that AI cannot be credited as an author because it lacks accountability was consistently observed across the policies of the four publishers reviewed, where publishers universally stated that AI bots must not be listed as authors.

“COPE's position that AI cannot be credited as an author because it lacks accountability (COPE, 2023) was consistently observed across the policies of the four publishers reviewed, where publishers universally stated that AI bots must not be listed as authors.”

• cites COPE, 2023Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

Sage states that authors are responsible for the submitted work and must verify the accuracy of any AI-generated content, correcting any errors.

“Sage reinforces this by stating that authors are responsible for the submitted work and must verify the accuracy of any AI-generated content, correcting any errors.”

Classified as probably not a substantial claim. May combine two claims.

Sage encourages authors to check sources and citations, including those generated by AI, cite AI-generated content appropriately, avoid plagiarism and copyright infringement, review content for AI biases, acknowledge AI limitations, check specific submission guidelines, and remain informed about ongoing debates regarding AI-generated content.

“Authors are also encouraged to check sources and citations, including those generated by AI, cite AI-generated content appropriately, avoid plagiarism and copyright infringement, review content for AI biases, acknowledge AI limitations, check specific submission guidelines, and remain informed about ongoing debates regarding AI-generated content.”

May combine two claims.

All four publishers articulate compliance expectations, but enforcement mechanisms remain notably vague across the board.

“All four publishers articulate compliance expectations, but enforcement mechanisms remain notably vague across the board.”

• AssumptionRests on an assumption: AI tools can be used to create fabricated submissions and to subvert peer review for commercial gain.
• AssumptionRests on an assumption: Current language models are not yet fully objective or factual.

May combine two claims.

Disclosure of AI use is a second area of broad convergence, though the specifics of what must be disclosed and where vary across publishers.

“Disclosure of AI use is a second area of broad convergence, though the specifics of what must be disclosed and where vary across publishers.”

May combine two claims.

Most publishers prohibited AI use in generating core research outputs, such as results or statistical analysis, but permitted its use for language editing and readability.

“This principle was echoed in the findings, where most publishers prohibited AI use in generating core research outputs, such as results or statistical analysis, but permitted its use for language editing and readability (Taylor and Francis, 2025; Sage, 2025).”

• cites Taylor and Francis, 2025Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Sage, 2025Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Emerald states that sensitive personal information must not be shared on AI platforms because AI providers collect such information for commercial purposes.

“Emerald adds that sensitive personal information must not be shared on AI platforms, since such information is collected for commercial purposes by AI providers.”

May combine two claims.

Taylor and Francis embraces AI as an opportunity for generating ideas and supporting non-native English-speaking authors, but this permissiveness does not extend to relieving authors of responsibility for what is ultimately submitted.

“Taylor and Francis, while more permissive in tone, embraces AI as an opportunity for generating ideas and supporting non-native English-speaking authors, but this permissiveness does not extend to relieving authors of the responsibility of what is ultimately submitted.”

May combine two claims.

Emerald requires all AI-created images to be flagged in the text and cited, and prohibits in-text reporting of statistics using AI, though AI may be used to aid analysis, provided this is disclosed.

“Emerald takes a more granular approach to transparency since it requires all AI-created images to be flagged in the text and cited, and that in-text reporting of statistics using AI be explicitly prohibited, though AI may be used to aid analysis, provided this is disclosed.”

May combine two claims.

Frontiers and Sage both mandate explicit disclosure when generative AI tools are used to generate text, images, or references.

“Frontiers and Sage both mandate explicit disclosure when generative AI tools are used to generate text, images, or references.”

May combine two claims.

Frontiers specifies that authors are accountable for the originality, validity, and integrity of their submissions, and that they must review AI-generated content and confirm its accuracy.

“Frontiers specifies that authors are accountable for the originality, validity, and integrity of their submissions, and that they must review AI-generated content and confirm its accuracy.”

May combine two claims.

Sage provides the most comprehensive guidance, requiring authors to verify AI-generated content for accuracy and hallucinations.

“Sage provides the most comprehensive guidance, requiring authors to verify AI-generated content for accuracy and hallucinations.”

Classified as probably not a substantial claim. May combine two claims.

Publishers acknowledge substantial risks associated with AI use in scholarly publishing, including factual errors, fabricated references, unattributed text, confidentiality breaches, and copyright concerns, particularly with AI-generated images.

“Publishers acknowledge substantial risks associated with AI use in scholarly publishing, including factual errors, fabricated references, unattributed text, confidentiality breaches, and copyright concerns, particularly with AI-generated images.”

May combine two claims.

The findings confirm that AI governance in scholarly publishing is both active and fragmented.

“The findings confirm that AI governance in scholarly publishing is both active and fragmented.”

• AssumptionRests on an assumption: The four publishers examined are representative of major scholarly publishers in AI governance.

May combine two claims.

All four publishers examined prohibit listing AI as an author and require human accountability for AI-assisted content.

“All four publishers examined prohibit listing AI as an author and require human accountability for AI-assisted content”

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

Publishers themselves identify potential benefits of AI, including improved accessibility for non-native English-speaking researchers, enhanced editorial screening, and streamlined matching.

“Publishers themselves identify potential benefits of AI, including improved accessibility for non-native English-speaking researchers, enhanced editorial screening, and streamlined matching.”

May combine two claims.

Across the four publishers examined, responsible, limited AI assistance, particularly for language editing and readability, is broadly accepted, provided authors retain intellectual control, verify accuracy, and comply with disclosure requirements.

“Across the four publishers examined, the study found that responsible, limited AI assistance, particularly for language editing and readability, is broadly accepted, provided authors retain intellectual control, verify accuracy, and comply with disclosure requirements.”

May combine two claims.

The four publishers diverge considerably in what they permit, how they frame disclosure obligations, and how they address enforcement.

“yet they diverge considerably in what they permit, how they frame disclosure obligations, and how they address enforcement.”

• AssumptionRests on an assumption: The four publishers examined are representative of major scholarly publishers in AI governance.

May combine two claims.

While there is broad agreement on core issues, such as prohibiting AI authorship, requiring disclosure, and maintaining human accountability, differences remain in areas like literature review support, image creation, and idea generation.

“While there is broad agreement on core issues, such as prohibiting AI authorship, requiring disclosure, and maintaining human accountability, differences remain in areas like literature review support, image creation, and idea generation.”

May combine two claims.

Uploading confidential materials such as manuscripts under review to external AI platforms is explicitly prohibited, as is listing AI as an author.

“Uploading confidential materials such as manuscripts under review to external AI platforms is explicitly prohibited, as is listing AI as an author.”

May combine two claims.

While all four publishers require disclosure of AI use, none of the reviewed policies articulate robust verification mechanisms to confirm that disclosure requirements are being met.

“While all four publishers require disclosure of AI use, none of the reviewed policies articulate robust verification mechanisms to confirm that disclosure requirements are being met.”

Classified as probably not a substantial claim. May combine two claims.

WAME recommendations state that authors should acknowledge the sources of their materials, be transparent when chatbots are used, and provide information about how they were used.

“Authors should acknowledge the sources of their materials, i.e., authors should be transparent when chatbots are used and provide information about how they were used.”

• cites World Association of Medical Editors, 2023Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing).

WAME recommendations state that editors and reviewers should specify any use of chatbots in the evaluation of the manuscript and generation of reviews and correspondence.

“Editors and reviewers should specify, to authors and each other, any use of chatbots in the evaluation of the manuscript and generation of reviews and correspondence.”

• cites World Association of Medical Editors, 2023Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing).

ICMJE states that chatbots should not be listed as authors since they are not responsible for the articles' originality, integrity, and accuracy.

“In addition, chatbots should not be listed as authors since they are not responsible for the articles' originality, integrity, and accuracy, which are requirements for authorship.”

• cites International Committee of Medical Journal Editors, 2025Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

ICMJE updated its recommendations in April 2025 to require explicit disclosure of AI use and to confirm that AI tools cannot qualify as authors because they cannot assume responsibility for published work.

“The International Committee of Medical Journal Editors (ICMJE) updated its recommendations in April 2025 to require explicit disclosure of AI use and to confirm that AI tools cannot qualify as authors because they cannot assume responsibility for published work (International Committee of Medical Journal Editors, 2025).”

• cites International Committee of Medical Journal Editors, 2025Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Classified as probably not a substantial claim.

The study was guided by four internationally recognized frameworks: the STM ethical and practical guidelines on AI use, the COPE position statement on authorship and AI tools, the WAME recommendations on chatbots and generative AI in scholarly publishing, and the ICMJE standards.

“The study was guided by the STM ethical and practical guidelines on AI use, COPE position statement on authorship and AI tools, WAME recommendations on chatbots and generative artificial intelligence in relation to scholarly publications, and the ICMJE.”

• cites STM, 2023Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites COPE, 2023Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites WAME, 2023Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing).

WAME recommendations state that editors need appropriate digital tools to deal with the effects of chatbots on publishing and help them detect content generated or altered by AI.

“Editors need appropriate digital tools to deal with the effects of chatbots on publishing and help them detect content generated or altered by AI.”

• cites World Association of Medical Editors, 2023Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

WAME recommendations state that authors must take public responsibility for their work, including the accuracy of what is presented and the absence of plagiarism.

“Authors must take public responsibility for their work since authors are responsible for the material provided by a chatbot in their paper, including the accuracy of what is presented and the absence of plagiarism.”

• cites World Association of Medical Editors, 2023Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

Bibliometric analysis of publisher instructions found that while the majority disallowed AI as an author and called for disclosure, actual guidance was inconsistent in scope, location, and required detail, suggesting burdens on authors and editors and a risk of policy confusion.

“Bibliometric analysis of publisher instructions found that while the majority disallowed AI as an author and called for disclosure, actual guidance was inconsistent in scope, location, and required detail, suggesting burdens on authors and editors and a risk of policy confusion (Ganjavi et al., 2023, 2024).”

• cites Ganjavi et al., 2023, 2024Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

WAME recommendations state that only humans can be authors, and chatbots cannot be authors.

“Only humans can be authors, and chatbots cannot be authors.”

• cites World Association of Medical Editors, 2023Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

ICMJE states that authors should not cite AI nor list it as a co-author and should avoid any form of plagiarism.

“It is further noted that authors should not cite AI nor list it as a co-author and should avoid any form of plagiarism.”

• cites International Committee of Medical Journal Editors, 2025Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Classified as probably not a substantial claim. May combine two claims.

Publishers were included if they maintained a publicly accessible AI-specific policy or had explicit AI-related clauses embedded in their author guidelines; operated at a global scale with multi-disciplinary coverage; and had updated their AI-related guidance within the preceding 2 years.

“Publishers were included if they maintained a publicly accessible AI-specific policy or had explicit AI-related clauses embedded in their author guidelines; operated at a global scale with multi-disciplinary coverage; and had updated their AI-related guidance within the preceding 2 years to reflect the generative AI developments since 2022.”

May combine two claims.

STM guidelines emphasize that while AI can be a supportive tool for refining, correcting, and editing manuscripts, its use should not extend to the manipulation of research data or attribution of authorship.

“The STM guidelines emphasize that while AI can be a supportive tool for refining, correcting, and editing manuscripts, its use should not extend to the manipulation of research data or attribution of authorship.”

• cites STM (2023)Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

These differences highlight that while the conceptual framework provides general ethical anchors, the interpretation at the publisher level depends on disciplinary priorities and publisher philosophies.

“These differences highlight that while the conceptual framework provides general ethical anchors, the interpretation at the publisher level depends on disciplinary priorities and publisher philosophies.”

Classified as probably not a substantial claim. May combine two claims.

These guidelines provide the normative basis for acceptable and ethical use of GenAI in scholarly publishing, particularly emphasizing disclosure, transparency, human accountability, and prohibition of AI as an author.

“These guidelines provide the normative basis for acceptable and ethical use of generative artificial intelligence (GenAI) in scholarly publishing, particularly emphasizing disclosure, transparency, human accountability, and the prohibition of AI as an author.”

• cites STM (2023)Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites COPE (2023)Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites World Association of Medical Editors (2023)Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

WAME's emphasis on authenticity and the risk of misinformation resonates with Emerald's restrictive policies, while Sage's more flexible guidelines reflect the balance between innovation and safeguarding research integrity.

“WAME (2023) emphasis on authenticity and the risk of misinformation resonates with Emerald's restrictive policies, while Sage's more flexible guidelines reflect the balance between innovation and safeguarding research integrity.”

• cites WAME (2023)Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Classified as probably not a substantial claim.

The extracted texts were analyzed using thematic analysis following the Braun and Clarke (2006) framework.

“The extracted texts were analyzed using thematic analysis, which involved coding, categorizing, and identifying recurring themes across the four publishers' documents. Following Braun and Clarke (2006) framework, the analysis proceeded through stages of familiarization, coding, theme generation, and interpretation.”

• cites Braun and Clarke (2006) sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

STM guidelines recommend that using publicly available GenAI as a basic tool that supports authors in refining, correcting, formatting, and editing texts and documents is permissible.

“Using publicly available GenAI as a basic tool that supports authors in refining, correcting, formatting, and editing texts and documents is permissible.”

• cites STM, 2023Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Classified as probably not a substantial claim.

The study adopted a qualitative research design appropriate for exploring meanings, expectations, and interpretations rather than measuring frequencies or testing hypotheses.

“This study adopted a qualitative research design, which is appropriate for exploring meanings, expectations, and interpretations rather than measuring frequencies or testing hypotheses (Lim, 2024).”

• cites Lim, 2024 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

The convergence on language assistance reflects a recognition that AI-powered language tools are now deeply embedded in academic writing practice globally.

“This convergence on language assistance reflects a recognition that AI-powered language tools are now deeply embedded in academic writing practice globally.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The assistive/generative distinction is a notable policy distinction as it acknowledges that not all AI use is equivalent and attempts to calibrate disclosure obligations accordingly.

“This assistive/generative distinction is a notable policy distinction as it acknowledges that not all AI use is equivalent and attempts to calibrate disclosure obligations accordingly.”

Probably not a substantial claim: may not pass S3 (load-bearing).

ICMJE recommends that authorship be based on four criteria: substantial contributions; drafting or critical review; final approval; and agreement to be accountable.

“The ICMJE helps in defining the roles of authors and contributors, and recommends that authorship be based on the following 4 criteria:”

• Also stated in: § International Committee of Medical Journal Editors – ICMJE · plain

  “Substantial contributions to the conception or design of the work; or the acquisition, analysis, or interpretation of data for the work; AND”

• Also stated in: § International Committee of Medical Journal Editors – ICMJE · plain

  “Agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved” (International Committee of Medical Journal Editors, 2025).”

• cites International Committee of Medical Journal Editors, 2025Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

ICMJE states that authors should review and edit their manuscripts before submission, as AI can generate incorrect or biased information.

“Therefore, authors should review and edit their manuscripts before submission, as AI can generate incorrect or biased information.”

• cites International Committee of Medical Journal Editors, 2025Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing).

STM guidelines recommend that GenAI cannot be used to create, alter, or manipulate original research data and results, such as images, blots, photographs, x-rays, and measurements.

“GenAI cannot be used to create, alter, or manipulate original research data and results, such as images, blots, photographs, x-rays, and measurements.”

• cites STM, 2023Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing).

STM guidelines recommend that GenAI cannot be credited as an author of a published work.

“GenAI cannot be credited as an author of a published work”

• cites STM, 2023Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Classified as probably not a substantial claim. May combine two claims.

The document analysis protocol included categories such as permitted AI uses, prohibited AI uses, disclosure requirements, authorship rules, treatment of AI-generated images/data, peer review guidance, and implementation mechanisms, ensuring systematic and consistent data extraction across all four publishers.

“The protocol included categories such as permitted AI uses, prohibited AI uses, disclosure requirements, authorship rules, treatment of AI-generated images/data, peer review guidance, and implementation mechanisms. This ensured systematic and consistent data extraction across all four publishers.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

COPE recommends that authors who utilize AI should disclose in the materials and methods section how they have used AI and indicate which tools were used.

“Therefore, it is recommended that authors who utilize AI should disclose in the materials and methods section how they have used AI and indicate which tools were used.”

• cites COPE, 2023Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing).

STM guidelines recommend that authors must disclose any use of GenAI that transcends basic use cases so an editorial decision can be made as to its legitimacy.

“Authors must disclose any use of GenAI that transcends those use cases so an editorial decision can be made as to its legitimacy.”

• cites STM, 2023Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

The findings of this study show how these global frameworks are reflected, adapted, or contested within individual publisher policies, highlighting both convergence and divergence in practice.

“The findings of this study show how these global frameworks are reflected, adapted, or contested within individual publisher policies, highlighting both convergence and divergence in practice.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

COPE acknowledges that AI tools cannot be listed as authors because they cannot meet authorship requirements, manage copyright and licensing agreements, or handle conflicts of interest.

“COPE acknowledges that the use of AI tools in research publications is growing, but these tools cannot be listed as authors (COPE, 2023). This is because AI tools cannot meet the requirements of authorship, as they are not responsible for any work submitted for publication. Among other things, AI tools cannot manage copyright and licensing agreements and cannot handle conflicts of interest.”

• cites COPE, 2023Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

Emerald's layered approach reflects a sophisticated attempt to distinguish between different functional uses of AI within a single manuscript.

“This layered approach reflects a sophisticated attempt to distinguish between different functional uses of AI within a single manuscript.”

May combine two claims.

COPE and WAME state that AI may assist, but it cannot assume authorship or accountability, and its use must be explicitly disclosed.

“Similar positions are echoed by the Committee on Publication Ethics (COPE, 2023) and the World Association of Medical Editors (WAME). Collectively, these statements set important norms whereby AI may assist, but it cannot assume authorship or accountability, and its use must be explicitly disclosed (World Association of Medical Editors, 2023).”

• cites COPE, 2023Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites World Association of Medical Editors, 2023Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Classified as probably not a substantial claim. May combine two claims.

ICMJE states that activities that do not qualify one to be an author include acquisition of funding; general supervision of a research group or general administrative support; and writing assistance, technical editing, language editing, and proofreading.

“It was stated that activities that do not qualify one to be an author include acquisition of funding; general supervision of a research group or general administrative support; and writing assistance, technical editing, language editing, and proofreading (International Committee of Medical Journal Editors, 2025).”

• cites International Committee of Medical Journal Editors, 2025Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim: may not pass S3 (load-bearing).

Inclusion and exclusion criteria were applied to ensure analytical rigor and consistency.

“Inclusion and exclusion criteria were applied to ensure analytical rigor and consistency.”

Classified as probably not a substantial claim. May combine two claims.

The stated benefits and risks are drawn directly from policy documents and represent publishers' own framing of the AI landscape, and they are not independently verified or empirically assessed outcomes of this study.

“The stated benefits and risks are drawn directly from policy documents and represent publishers' own framing of the AI landscape, and they are not independently verified or empirically assessed outcomes of this study.”

Classified as probably not a substantial claim. May combine two claims.

These findings are grounded in the documentary analysis of publicly available publisher policies and should be understood as a mapping of stated expectations, not as evidence of actual author behavior, compliance rates, or measurable outcomes.

“These findings are grounded in the documentary analysis of publicly available publisher policies and should be understood as a mapping of stated expectations, not as evidence of actual author behavior, compliance rates, or measurable outcomes.”

Probably not a substantial claim: may not pass S3 (load-bearing).

This tension between enabling technological innovation and maintaining ethical safeguards has been noted in recent scholarship on AI in publishing.

“This tension between enabling technological innovation and maintaining ethical safeguards has been noted in recent scholarship on AI in publishing (Clark, 2025; Pongrac, 2024).”

• cites Clark, 2025 sourceNot checked — no full text of the source
• cites Pongrac, 2024Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

May combine two claims.

The conceptual framework provided by international editorial bodies serves as the ethical compass guiding AI use in scholarly publishing, while the findings demonstrate how these principles are translated into heterogeneous publisher policies.

“In sum, the conceptual framework provided by international editorial bodies serves as the ethical compass guiding AI use in scholarly publishing, while the findings demonstrate how these principles are translated into heterogeneous publisher policies.”

Probably not a substantial claim: may not pass S2 (checkable).

The expectation that authors personally verify all AI-generated content for accuracy, correct hallucinations, and navigate rapidly changing policies across publishing places a disproportionate burden on researchers who may lack institutional support, advanced English proficiency, or familiarity with evolving AI tools.

“The expectation that authors personally verify all AI-generated content for accuracy, correct hallucinations, and navigate rapidly changing policies across publishing places a disproportionate burden on researchers who may lack institutional support, advanced English proficiency, or familiarity with evolving AI tools.”

Classified as probably not a substantial claim.

The study does not empirically assess whether these policies reduce policy breaches, enhance trust, or improve compliance in practice.

“The study does not empirically assess whether these policies reduce policy breaches, enhance trust, or improve compliance in practice.”

May combine two claims.

These divergences reflect the evolving nature of AI governance in scholarly publishing, where global principles are being locally interpreted by publishers.

“These divergences reflect the evolving nature of AI governance in scholarly publishing, where global principles are being locally interpreted by publishers.”

Probably not a substantial claim: may not pass S2 (checkable).

The gap between policy articulation and enforcement infrastructure represents a meaningful limitation of the current governance landscape that warrants further attention from publishing associations and institutional policymakers.

“The gap between policy articulation and enforcement infrastructure represents a meaningful limitation of the current governance landscape that warrants further attention from publishing associations and institutional policymakers.”

Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

Taylor and Francis's more permissive stance toward AI-assisted language improvement reflects an awareness that non-native English-speaking researchers face distinct challenges.

“Taylor and Francis's more permissive stance toward AI-assisted language improvement reflects an awareness that non-native English-speaking researchers face distinct challenges.”