AI Policy

Summary

Authors may use generative AI (GAI) when preparing their submissions but are required to be transparent about its use. Authors remain fully responsible for their submission and for complying with all relevant policies and research integrity standards.

PC members and subreviewers are prohibited from using GAI during the review process.

Detailed policy for authors

Authors may use generative AI where it genuinely assists their work; the following are some acceptable use cases: for instance, authors may use GAI as a programming assistant, to polish text that was originally written by the authors, or to generate code that creates scientific plots from experimental data.

Authors must be transparent about the use of GAI when preparing their manuscripts. Unless obvious in the main body of the manuscript, e.g., since the subject of the research investigates AI in graph drawing, authors must declare their GAI use (if any) in a short statement on the title page of the manuscript, which does not count towards the permitted line limit. Failure to do so is a reason for rejection.

Authors are always ultimately responsible for the content of the submission. The use of GAI carries several risks and limitations that may lead to rejection of a submission and that the authors should be aware of, e.g., plagiarism, model biases, false or fabricated information, erroneous code, and incorrect citation of sources/attribution of original ideas. Authors should review the output generated by GAI for accuracy, bias, integrity, and completeness and add the appropriate citations. These author responsibilities extend beyond the review process: failure to comply with research integrity standards, including undisclosed GAI use or fabricated content, may result in post-publication consequences such as retraction or other corrective action.

Authors should further be aware of the risks of using GAI in relation to the sharing of confidential information, such as personal data and intellectual property. Information entered into a GAI application, including text, data, and prompts, may be stored by the GAI application and reused for other purposes, such as training the underlying language model or improving its chat functionality. As a result, confidential information may be leaked to other GAI users.

When using GAI, authors should respect the applicable institutional, national, and international laws, research integrity standards and ethics guidelines, as well as privacy and data protection regulations such as the EU General Data Protection Regulation (GDPR).

Detailed policy for reviewers

GAI may not be used in any manner when writing review reports. This strict ban comes for multiple reasons:

• The peer-review process of GD is confidential. Hence, the contents of submitted manuscripts, as well as the reviewers’ reports and PC discussions judging the value of these submissions, constitute confidential information. Entering such information into third-party generative AI systems, even for polishing text in the reviewer’s report, is regarded as a breach of confidentiality.
• Further, PC members and subreviewers are selected based on their technical expertise and knowledge of the field; hence, they may not delegate their judgment of a submission to a GAI system.
• Finally, the use of GAI also has other significant drawbacks, including the risk of bias and the production of incomplete, fabricated, or erroneous outputs. The exclusion of GAI from the reviewing process promotes the non-discriminatory, transparent, and independent evaluation of submissions.

Due to the confidentiality requirements, this policy is asymmetric, and reviewers should be aware that authors are not prohibited from using GAI to prepare their submissions. Authors are, however, required to declare any GAI use.

PC members and subreviewers are requested to read the AI policy of GD carefully before reviewing the manuscripts.