Responsible AI Use Policy

Preamble

JOTMI publishes research on technology, management and innovation at a time when research itself is undergoing an unprecedented technological transformation. The arrival of large language models (LLMs) and generative artificial intelligence tools has irreversibly altered the practices of scholarly writing, analysis, peer review and scientific communication. To ignore this shift would be a disservice to our readers; to react with blanket prohibitions would be naïve and counterproductive.

This policy takes a deliberate middle position: artificial intelligence is a legitimate, useful technology that, when properly employed, benefits the academic ecosystem. Its use is not, in itself, an editorial breach. What this journal requires —and verifies— is transparency, substantive human oversight, bibliographic integrity and empirical truthfulness. Authorship, together with the scientific and ethical responsibility for every manuscript, remains entirely within the human domain.

This document complements and is to be read jointly with the JOTMI Publication Integrity Protocol and Extended Retraction Policy, of which it constitutes the preventive and orientative counterpart.

1. Authors

General stance. JOTMI does not reject or penalise a manuscript merely because it was supported by artificial intelligence. What is evaluated is the quality, originality and honesty of the work, not the tool with which it was produced.

Compatible uses that do not require detailed disclosure. Purely instrumental uses, with low influence on substantive content, do not require detailed disclosure: spelling and grammar correction, translation of the text from the author's native language, lexical and stylistic suggestions, bibliographic formatting, file-format conversion and audio transcription. Their mention in the declaration is left to the author's discretion.

Uses requiring explicit disclosure. When the intervention of artificial intelligence affects substantive content, it must be specifically declared. This includes, among others:

  • the generation or rewriting of fragments of the text, beyond linguistic correction;
  • the generation, in whole or in part, of the abstract;
  • assistance in formulating hypotheses, designing the methodology or interpreting results;
  • the production of statistical-analysis code, data-processing scripts or database queries;
  • the analysis of primary data by means of AI (automated classification, qualitative coding, topic modelling);
  • bibliographic search delegated to AI systems, particularly where references suggested by the model have been incorporated;
  • the generation of figures, schematics, diagrams or images.

The declaration must specify the tool used, the sections affected, the purpose and the level of subsequent human oversight. This information is added to the editorial file and does not, by itself, determine the outcome of the evaluation, except where it reveals one of the prohibited uses.

Form and placement of the declaration. The declaration appears in a section titled "Declaration on the Use of Artificial Intelligence", placed after the Acknowledgements and before the References; in manuscripts undergoing double-blind review, after the Conclusions. Where the use does not exceed the compatible cases, the following statement suffices: "The authors did not use generative artificial intelligence tools in the substantive preparation of this manuscript."

Prohibited uses. The following practices constitute a serious breach —entailing rejection of the manuscript or, if it has already been published, its retraction:

  • presenting as original any content generated by AI without verification or substantive rewriting;
  • attributing authorship, in whole or in part, to an AI system;
  • including false references, in any of their forms:
    • fabricated or non-existent: they have no real basis whatsoever;
    • ghost or fraudulent: the identifier (DOI, link) exists but resolves to a different document, or the source does not support what is attributed to it;
    • orphan: cited in the text with no corresponding entry in the reference list, or vice versa;
    • with an invalid DOI or one that points to a retracted work;
  • fabricating data, results, observations or transcripts by means of AI;
  • inserting hidden text, microtext or residual instructions (prompts) intended to manipulate the editorial process or subsequent indexing;
  • substituting Latin characters with homoglyphs from other alphabets (Cyrillic, Greek) in order to evade similarity-detection systems;
  • artificially inflating the reference list with citations bearing no substantive relation to the text (excessive self-citation or courtesy citations);
  • engaging in practices characteristic of paper mills: the sale or purchase of authorship, hyper-prolific publication without verifiable contribution, or impersonation of academic identity.

Final responsibility. Signature of the manuscript implies that each co-author has read the complete content and personally verified every cited reference prior to submission. This verification is neither transferred to any automated tool nor diminished by its use.

2. Peer reviewers

Peer review is an intellectual function whose delegation to AI systems, without disclosure or oversight, contravenes editorial trust and compromises the confidentiality of the manuscript.

The use of AI for ancillary tasks is permitted —verifying references, translating the manuscript into the reviewer's working language, organising comments— provided that: (a) it is declared to the handling editor in the confidential-comments section; (b) the complete manuscript or substantial parts of it are not uploaded to services that do not guarantee confidentiality; (c) the judgement on contribution, methodology and relevance comes from the reviewer and not from the model.

The journal attends to certain signals of inappropriate delegation: response times close to zero, language characteristic of automated generation, generic comments unrelated to the empirical content of the manuscript, or recurring patterns across the reports of a single reviewer. Such signals prompt a private enquiry, with no automatic sanction; nonetheless, inappropriate and systematic use may lead to removal from the panel.

JOTMI also reserves the use of internal AI-based editorial-assistance tools (reviewer suggestion, technical pre-evaluation, bibliographic verification), subject to the same requirement of transparency, which is reported on the journal's institutional website.

3. Editors and editorial staff

Editors and technical staff are bound by the same requirements of transparency and human oversight. Substantive decisions —to accept, request revisions, reject or retract— are non-delegable human responsibilities: the audit tools provide technical evidence but do not issue decisions. An internal record is kept of consultations to AI systems that have influenced a decision, available to the Editorial Board and to appeal procedures.

4. Editorial verification infrastructure

In line with the standard practice of indexed journals, JOTMI uses commercial similarity- and plagiarism-detection tools (iThenticate and equivalents). In a complementary manner, it operates its own audit infrastructure, directed at the dimensions that similarity detection does not cover. It currently comprises three modules:

  • AAM (Authorship, Affiliation and AI detection): verifies authorial identity, the consistency of affiliations and ORCID traceability; detects patterns consistent with paper mills and file manipulation (homoglyphs, hidden text); and analyses signals of AI-generated text without curation (workslop).
  • BIM (Bibliographic Integrity): checks the existence and relevance of each reference; detects fabricated, ghost and orphan references, invalid DOIs or those corresponding to retracted works, hijacked journals and the currency of the bibliography.
  • DSRM (Data and Statistics): verifies the coherence between the data described, the methods declared and the results reported, preventing evaluations based on non-existent data.

The detection of workslop was previously carried out in a separate module, WAM, subsequently integrated into AAM. The modules may be applied at any stage of the process, including after publication, and their results constitute one further technical input to the human editorial decision.

5. Consequences regime

Breaches are graded according to their severity and their deliberate or negligent character:

  • Minor (omission of a minor disclosure; template declarations providing no real information): resolved through a request for correction during the process, without affecting the decision.
  • Moderate (invalid references not detected by the author; AI-generated sections without disclosure but verifiable as substantive): give rise to a major revision or rejection, at the discretion of the handling editor.
  • Serious (data fabrication, false references, evasion techniques, attribution of authorship to AI, paper mills): the manuscript does not advance to peer review or is rejected during the editorial process. Retraction is exceptional: it applies only where the breach is detected once the article has been published, in which case the Retraction Policy is activated, with formal retraction and a permanent mark on the document.

6. Confidentiality and official communications

The entire editorial process is confidential in character. Manuscripts, internal observations and audit results constitute reserved information: they are not disclosed, published, commented on social media or used for purposes of exposure. The audit is an instrument serving the improvement of the work and the integrity of the scientific record, not a denunciation mechanism; this confidentiality applies equally to rejected and to accepted manuscripts.

Fraud warning. Three rules should be borne in mind:

  • The only official email domains for JOTMI editorial communications are @jotmi.org and @uahurtado.cl (Universidad Alberto Hurtado, host institution). Communications from other domains claiming to represent the journal —including personal accounts, visually similar domains or social networks— are not authentic and must be reported to the editorial office.
  • JOTMI does not charge fees for processing manuscripts, does not request deposits or transfers, and does not condition the speed of review on any payment whatsoever. Any request for payment must be treated as an attempted fraud.
  • Special issues originate exclusively from the initiative of the Editorial Board and are announced on jotmi.org. The journal does not accept external proposals to coordinate special issues, nor invitations to act as a guest editor received through informal channels; faced with such an invitation, it should be verified against the official site and, if it does not appear there, treated as spurious.

7. Use of JOTMI content by external AI systems

JOTMI content is published under open-access licences that permit its reading, distribution and reuse with attribution. The use of that content for training AI systems is a distinct matter: the journal reserves the right to apply reasonable technical measures (robots.txt configuration, server headers) to regulate massive automated scraping, without affecting readers' access.

8. Validity and review

This policy enters into force on its date of publication and applies to manuscripts submitted from that moment onward. The consequences regime for serious breaches additionally applies to the journal's historical archive, as part of post-publication audit. The policy is reviewed at least once per academic year, in line with technological developments and the evolution of international standards (COPE, ICMJE, WAME).

Institutional document. Complementary to the JOTMI Publication Integrity Protocol and Extended Retraction Policy.