Why this policy exists
Journals, publishers and ethics bodies have set out expectations about the use of generative artificial intelligence in research. The International Committee of Medical Journal Editors and the Committee on Publication Ethics have both stated that such tools cannot be listed as authors, because authorship carries accountability that software cannot take, and that authors should disclose their use of such tools as the journal requires. These tools also fail in specific ways that matter in evidence synthesis: they can invent studies, misstate numbers and produce references that look real and are not.
Clients are entitled to know how their material is handled and which tools are used on it. This policy is written to answer that question.
Principles that apply in every case
- People are responsible. Every analysis, statement and reference in work we deliver is checked by a person who takes responsibility for it. No tool is treated as an author or as an authority.
- Data and citations are never generated. We do not use any tool to create, fill in or estimate data, study results or references. Every reference is checked against its source, and every number is traced to its origin.
- Confidential material stays confidential. Unpublished manuscripts, data and protocols are not entered into external tools that retain or learn from what they receive, unless the client has agreed in writing.
- Disclosure follows the journal. Where a journal requires authors to declare the use of AI tools, we will say, on request, whether and how any such tool was used on a project, so that the authors can make an accurate declaration.
- Clients decide for their own work. Authors are responsible for following the AI policy of the target journal in their own writing.
Our position on AI tools
[OWNER DECISION REQUIRED] The two options below describe different practices. Keep only the one that is true for how work is actually done, and delete the other and this notice. The statement must match the real workflow, including how this website's content was prepared.
Option A: no generative AI used on client work
We do not use generative artificial intelligence tools to write, analyze or edit client material. The analysis is carried out with statistical software, and the text is written and edited by people. Conventional tools such as spelling and grammar checkers, reference managers and statistical packages are used as usual and are not regarded as generative AI.
Option B: limited use under human control
We may use AI-assisted tools for limited tasks, such as language suggestions or the formatting of text that does not contain confidential data, and never for generating data, results, interpretations or references. Any use is reviewed in full by a person who is responsible for the output. Client material is not entered into such tools without the client's written agreement, and use on a project is disclosed to the client on request.
What we ask of clients
Please check the policy of your target journal on the use of artificial intelligence and follow it in your own writing. If you have used an AI tool to draft or edit text before sending it to us, tell us, so that the material can be checked with that in mind and so that you can make the declaration the journal asks for. Please do not send us text that contains references you have not verified.
Questions and changes
If you have a question about how a tool was used on your project, or a concern about this policy, please write to us through the contact page. This policy is reviewed as journal and publisher requirements change, and the date of the latest update is shown at the foot of the page. It sits alongside the research integrity and authorship statement.
References
- International Committee of Medical Journal Editors. Recommendations for the conduct, reporting, editing, and publication of scholarly work in medical journals. Section on artificial intelligence-assisted technology. Available at icmje.org (check the current version).
- Committee on Publication Ethics. Authorship and AI tools: COPE position statement. Available at publicationethics.org.