Charte relative à l’usage de l’intelligence artificielle (version française)
Le Rubicon recognizes that artificial intelligence (AI) tools, and generative AI in particular, can be useful at various stages of preparing, reviewing, and editing an article. However, their use must adhere to three guiding principles: human responsibility, transparency, and confidentiality.
Human Responsibility. An AI tool cannot replace the judgment of the author, reviewer, or editorial team. The person using such a tool remains fully responsible for the resulting content, analyses, and decisions. An AI tool cannot be considered an author or co-author, nor can it be considered a source. Any information, quote, or references it suggests must be located, consulted, and verified in the original sources.
Transparency. Purely spelling, grammatical, or typographical corrections need not be disclosed. However, any use that has substantially produced or transformed published content, or that has contributed to the underlying intellectual work, must be disclosed in accordance with the procedures specified below.
Confidentiality. Articles submitted to Rubicon remain unpublished documents until they are posted online. Authors, reviewers, and members of the editorial team must take care not to share them in any way that could compromise this confidentiality.
- Authors
Authors must inform the Rubicon editorial team of any substantial use of an AI tool in preparing or drafting their article, and describe such use transparently. Authors must also comply with the platform’s editorial standards.
Authors remain fully responsible for the content of their text, its accuracy, rigour, and, where applicable, scientific quality, including for sections written with AI assistance. In particular, authors must ensure that their article contains no factual errors, fabricated quotes or references, plagiarism, or copyright infringement.
Uses limited to spelling, grammatical, or typographical corrections do not require disclosure. Substantial uses must be disclosed, particularly when an AI tool has been used to:
- substantially rephrase or translate all or part of a text ;
- generate or explore ideas ;
- help structure an argument or an article ;
- assist with a literature review or the classification of references ;
- assist with coding ;
- contribute to data analysis or visualization ;
- create an illustration, infographic, or other visual content.
When an AI tool is used to identify, classify, or suggest references, the author must locate and verify them in the original sources before citing them.
Any image generated or substantially modified by AI must be explicitly identified as such. It must not be presented as a photograph, an archival image, or any other authentic document. The author must also ensure that its use complies with applicable copyright and image-rights rules, and, more generally, that it does not misrepresent reality.
Whenever AI has been used to a substantial extent, the author must include a statement at the end of their text, following their brief biography:
Statement Regarding the Use of Artificial Intelligence. “The author declares having used [tool, version if known] for [purpose]. This use concerned [relevant sections or stages]. The author has verified the relevant content, information, citations, and references and assumes full responsibility for the published text.”
The use of AI in accordance with this charter does not affect the peer review process and does not constitute grounds for rejecting the article.
- Reviewers
Reviewers may use AI tools to assist with their evaluation work, provided that such use remains transparent and is carried out under their full responsibility.
Under no circumstances may AI replace the reviewer’s critical judgment or, on its own, produce an evaluation or editorial recommendation.
Reviewers must not upload a submitted, unpublished article in its entirety, or any accompanying material (appendices, images, data, or metadata) to an AI tool, unless the editorial team and the author have expressly authorized it and the tool offers appropriate confidentiality safeguards. They may, however, use an AI tool on limited sections of the manuscript or for supplementary research, provided they comply with this policy.
Using an AI tool solely to correct the language or formatting of the review report does not need to be disclosed, as long as it does not affect the content of the review. Any substantial use, however, must be disclosed to the editorial team. This includes, in particular:
- assistance in structuring the evaluation ;
- identifying points requiring attention ;
- searching the literature for contextual information ;
- summarizing or analyzing passages from the manuscript.
The disclosure must specify the tool used, its version (if known), and the exact nature of the assistance provided.
The mere use of an AI tool does not constitute grounds for excluding a review from the editorial process. However, a review may be excluded if such use compromised the confidentiality of the manuscript, if it was not disclosed when it should have been, or if it appears that the reviewer’s judgment was substantially delegated to the tool.
Reviewers who suspect substantial, undeclared use of AI by an author are encouraged to inform the editorial team. Such a suspicion does not, in and of itself, constitute proof of a violation, and no violation may be established solely on the basis of an AI-content detection tool.
III. Editorial Team
The Rubicon editorial team applies the same principles of accountability, transparency, and confidentiality to its own use of AI.
The team may use AI or language assistance tools at various stages of the editorial process, in particular for:
- translation ;
- linguistic, typographical, or stylistic editing ;
- preparing or adapting supplementary content ;
- creating summaries intended primarily for search engine optimization (SEO).
These tools do not replace the editorial team’s judgment; the team remains responsible for the changes made to articles and for publication decisions.
The editorial team must not upload a submitted but unpublished article in its entirety to an external AI tool without the author’s consent, unless the tool provides appropriate confidentiality guarantees and its use complies with Le Rubicon’s editorial practices.
As of the date of publication of this charter, Rubicon uses, in particular:
- DeepL, for translating articles ;
- Grammarly and Antidote, for language editing ;
- Mistral and Euria, in particular for generating SEO-optimized summaries once an article has been accepted for publication.
This list is illustrative and subject to change.
- Violations and Questions of Interpretation
In case of doubt about the scope of this policy or the need to disclose a particular use of AI, authors and reviewers are encouraged to consult the editorial team.
In case of doubt or non-compliance with this policy, the editorial team may request an explanation or additional disclosure and decide on appropriate measures according to the nature and seriousness of the violation. These measures may include a request to correct or complete the disclosure, a revision of the text, or, for an article that has already been published, the addition of a note or an erratum. In the most serious cases, they may extend to retraction of the article.
These decisions take into account, in particular, the significance of the practice in question, its potential impact on the integrity of the content, compliance with the rules on transparency and confidentiality, and whether the violation was intentional.
Publication date: September 17, 2026