03 · What You Need to Know
Why Academic Authorship Cannot Be Assigned to an AI System
What Does Academic Authorship Actually Mean?
Academic authorship identifies individuals who have made qualifying scholarly contributions and accept responsibility for the resulting work.
Being named as an author can carry professional recognition, career benefits, and intellectual credit. It also creates expectations about the author's ability to explain, defend, and correct the published research.
Authorship therefore has at least two connected dimensions: contribution and accountability.
Scholarly contribution
Work that meaningfully contributes to developing, conducting, analyzing, interpreting, or communicating research.
Authorship accountability
Responsibility for the integrity of the published work, including appropriate responses to questions, errors, and concerns about research conduct.
A contribution may be substantial without automatically qualifying its source for authorship.
For example, a statistician, research assistant, language editor, or technical specialist may contribute to a manuscript, but whether a human contributor qualifies for authorship depends on the relevant disciplinary and journal criteria.
AI systems introduce a different issue. Even when their outputs contribute materially to a paper, they cannot undertake the responsibilities that major publishing policies attach to authorship.
What Does the ICMJE Say About AI Authorship?
The International Committee of Medical Journal Editors provides widely used recommendations for authorship in medical and related scholarly publishing.
Its authorship framework specifies four criteria, all of which must be satisfied by an individual recognized as an author.
- Making substantial contributions to the conception or design of the work, or to the acquisition, analysis, or interpretation of data.
- Drafting the work or critically reviewing it for important intellectual content.
- Giving final approval to the version that will be published.
- Agreeing to be accountable for all aspects of the work, including ensuring that questions concerning accuracy or integrity are appropriately investigated and resolved.
The ICMJE explicitly states that AI-assisted technologies should not be listed as authors because they cannot accept responsibility for the accuracy, integrity, and originality of submitted work.
Its recommendations also require human authors to review AI-generated material carefully and remain responsible for content produced with AI assistance.
The four criteria are ICMJE recommendations rather than universal legal requirements for every academic discipline. Nevertheless, their emphasis on accountability explains why AI-generated contributions do not automatically confer authorship.
The question of whether AI systems can satisfy established authorship criteria requires distinguishing observable task performance from the capacity to accept scholarly responsibility.
What Is COPE's Position on AI as an Author?
The Committee on Publication Ethics (COPE) has issued an explicit position on authorship and AI tools.
COPE states that AI systems cannot be listed as authors because they cannot take responsibility for submitted work.
Its position also identifies additional limitations: AI systems are not legal entities capable of declaring conflicts of interest or managing copyright and licensing agreements in the manner required of authors.
COPE further emphasizes that researchers who use AI tools for manuscript writing, image production, or research activities must be transparent about that use.
The human authors remain responsible for the manuscript, including material produced with AI assistance.
COPE's position does not mean that AI tools can never contribute to research. Rather, it establishes that contribution alone does not satisfy the responsibilities associated with formal authorship.
What Does Springer Nature Require?
Springer Nature's official AI guidance states that large language models do not meet its authorship criteria because they cannot take accountability for the work.
Consequently, an AI system should not appear in the author byline as though it were a human researcher.
Springer Nature also provides instructions for documenting AI assistance. The appropriate reporting location and requirements depend on the publication type and the nature of the assistance.
For example, its guidance addresses documentation of generative AI use in manuscripts and distinguishes certain forms of AI-assisted copyediting from other activities.
Researchers should consult the current policy applicable to their journal rather than assuming that all Springer Nature publications or other publishers impose identical disclosure procedures.
The important distinction is consistent: AI may be used within applicable restrictions, but accountability for the published work remains with human authors.
How Do Major Authorship Positions Compare?
| Authority |
Can AI Be an Author? |
Principal Reason |
Practical Requirement |
| ICMJE |
No |
AI cannot accept responsibility for accuracy, integrity, and originality. |
Human accountability and appropriate disclosure of AI assistance. |
| COPE |
No |
AI cannot assume authorship responsibility or fulfill relevant legal and ethical obligations. |
Transparency about AI use and responsibility retained by human authors. |
| Springer Nature |
No |
Large language models cannot take accountability for the work. |
Documentation of AI use according to the applicable publication policy. |
These positions are broadly aligned on the authorship question, although their detailed disclosure instructions are not identical.
Researchers should distinguish the general principle of human accountability from the specific reporting requirements imposed by a particular journal or publisher.
Why Can't an AI System Accept Academic Responsibility?
Academic authors are expected to respond to questions about their research, explain methodological decisions, address concerns about accuracy, and participate in corrections when necessary.
They may also need to make declarations concerning conflicts of interest, research ethics, copyright, and other publication requirements.
An AI system can generate explanations and assist with preparing responses. However, generating a response is not equivalent to assuming responsibility for the underlying research.
A language model does not independently undertake the legal and professional obligations that attach to a named human author.
Nor can a journal hold an AI system accountable through ordinary institutional, professional, or editorial procedures in the same way it can hold a human researcher accountable.
These limitations explain why an AI system cannot take responsibility for academic authorship under prevailing publishing frameworks.
What If AI Writes Most of the Manuscript?
The amount of text produced by AI does not automatically determine who qualifies as an author.
Suppose a researcher uses a language model to generate an extensive manuscript draft. The researcher then checks the sources, revises the arguments, verifies the analyses, and prepares the paper for submission.
The AI system still does not qualify for authorship under the policies discussed above.
However, this does not automatically mean that the manuscript is acceptable for publication.
The researcher must determine whether the extent of AI assistance complies with the journal's policy and whether the human contributors have made the intellectual contributions and accepted the responsibilities required for authorship.
Substantial AI drafting may also create risks involving unsupported claims, inaccurate citations, or inadequate scholarly engagement.
The appropriate question is who can legitimately claim authorship when AI produces much of a manuscript, not whether the system deserves a share of the author byline.
Can AI Be Named as a Coauthor if a Human Researcher Takes Responsibility?
Under the policies of ICMJE, COPE, and Springer Nature, no.
A human researcher's willingness to accept responsibility does not transform an AI system into an eligible author.
Authorship criteria apply to each named author, not merely to the author group collectively.
For example, under the ICMJE framework, every listed author must meet all four authorship criteria.
A human coauthor cannot satisfy the AI system's accountability requirement on its behalf while still treating the system as an independent author.
The appropriate alternative is to recognize the contribution through disclosure or acknowledgment, as required by the publication policy.
Can an AI System Be Listed as the Corresponding Author?
No, not under the major publishing policies discussed here.
The corresponding author has responsibilities extending beyond manuscript preparation. These typically include communication with the journal, coordination of required documentation, and responses to editorial questions.
Depending on the journal, the corresponding author may also be responsible for coordinating revisions and communicating with coauthors after publication.
An AI system may help a human researcher draft correspondence, but it cannot independently assume the formal responsibilities of the corresponding author.
The corresponding author should therefore be an eligible human author who can fulfill the journal's requirements.
Can AI Be Credited Without Being Listed as an Author?
Yes. Many publishing policies distinguish between authorship and acknowledgment or disclosure of assistance.
For example, the ICMJE recommends reporting AI writing assistance in the acknowledgment section and describing AI use for data collection, analysis, or figure generation in the methods section, as applicable.
COPE also requires transparency about the AI tool and how it was used, although the precise location of the disclosure should be checked against the relevant journal's instructions.
Such reporting recognizes the involvement of AI without assigning it the responsibilities of a human author.
Researchers should distinguish between acknowledging an AI tool and granting it formal academic authorship.
The former may be appropriate or required. The latter is not permitted under the major policies considered here.
Can AI Be Included in an Author Contribution Statement?
Author contribution statements generally describe the roles performed by individuals recognized as authors.
Some journals use the Contributor Roles Taxonomy (CRediT) to specify activities such as conceptualization, data curation, formal analysis, and writing.
AI systems may assist with tasks corresponding to some of these activities, but that does not make them eligible authors.
Researchers should follow the journal's instructions for reporting AI-assisted activities rather than inserting the AI system into a human author contribution statement as a coauthor.
Where AI assistance materially affected the research process, its role may need to be described in the methods or another designated disclosure section.
The distinction between AI assistance and formal author contributions is important because contribution taxonomies do not override authorship eligibility requirements.
Does Using AI Reduce the Human Authors' Responsibility?
No. Major publishing policies place responsibility for submitted content on the human authors, including material developed with AI assistance.
Researchers remain responsible for checking references, verifying claims, ensuring appropriate attribution, and evaluating the accuracy of any AI-assisted analysis.
An AI system's involvement does not excuse fabricated citations, unsupported interpretations, or errors in the manuscript.
Nor does it allow researchers to transfer responsibility for problematic content to the software provider.
Human authors must remain capable of explaining and defending the submitted research.
This principle is especially relevant to the responsibilities retained by researchers who use extensive AI assistance.
Could AI Authorship Policies Change in the Future?
Publishing policies may evolve as AI capabilities, legal frameworks, and scholarly practices change.
Debates about machine agency, intellectual contribution, and responsibility may also continue in research ethics and philosophy.
However, speculative future developments should not be confused with current publication requirements.
Under the major policies discussed here, AI systems cannot be listed as authors because they cannot assume the obligations required for academic authorship.
Researchers should verify the current instructions of their intended journal before submission, particularly when publisher policies have recently been updated.
