AI Policy

At CJAIM, we recognize that Artificial Intelligence is a transformative scientific instrument. Our policy is rooted in transparency and human accountability. We encourage authors to use AI to improve the clarity, reach, and depth of their research, provided these tools are used ethically and documented with precision.

AUTHORSHIP AND ACCOUNTABILITY

In accordance with COPE and ICMJE guidelines, AI tools (e.g., ChatGPT, Claude, specialized medical LLMs) cannot be listed as authors. Authorship requires legal and ethical accountability that AI cannot fulfill.

Authors are fully responsible for the accuracy of their work. This includes verifying all medical dosages, clinical guidelines, and citations generated or processed by AI.

TIERED DISCLOSURE FRAMEWORK

CJAIM distinguishes between administrative assistance and methodological contribution.

Tier 1: Linguistic assistance

Description: Using AI for grammar, spelling, style, translation, or light rewording of non-native English text. Disclosure requirement: Disclose in the Acknowledgements - tool name, version, and date of use.

Tier 2: Research assistance

Description: Using AI for literature search and summarization, brainstorming, structural outlining, or coding help. Disclosure requirement: Disclose in the Methods - tool name, version, prompts or prompt categories, the verification procedure, and which human author is responsible.

Tier 3: Methodological assistance

Description: Using AI for data analysis, coding scripts (R/Python), identifying patterns in clinical data, image or figure generation, or synthetic data. Disclosure requirement: Disclose in the Methods with full prompt transparency, model specificity, the validation procedure, and data-privacy confirmation. Synthetic data must be clearly labelled.

PROGRESSIVE REQUIREMENTS FOR METHODOLOGICAL AI

If AI was used as a core component of the research process (Tier 3), CJAIM requires:

  • Prompt Transparency: Authors should be prepared to provide the prompts used to generate results or analyze data. We encourage including these in the Supplementary Material.
  • Model Specificity: Always state the model name and version (e.g., "Statistical analysis was optimized using GPT-4o with a custom Python script for Bayesian inference").
  • Data Privacy: Authors must certify that no un-anonymized patient data was uploaded to public or non-secure AI models.

AI IN PEER REVIEW

  • We encourage authors to use AI to check their manuscripts against CONSORT or PRISMA checklists prior to submission to speed up the review cycle.
  • AI must not be used to peer review a manuscript. A manuscript under review is a confidential document, and entering it into an external model breaches that confidentiality. This applies to reviewers and to editors alike.
  • Where the journal runs an automated check of its own (for example a similarity report), the finding is verified by a human editor before it is acted on, and every editorial decision is made by a human.

AI USAGE DECLARATION TEMPLATE

Include this declaration whenever an AI tool was used - in the Acknowledgements for Tier 1, in the Methods for Tier 2 and Tier 3:

"During the preparation of this work, the author(s) used [NAME OF TOOL/SERVICE] in order to [REASON]. After using this tool, the author(s) reviewed and edited the content as needed and take(s) full responsibility for the final scientific output. The specific prompts used are available in [SUPPLEMENTARY FILE / UPON REQUEST]."