Artificial Intelligence in Scientific Figure and Graphical Abstract Creation: Declaration Requirements, Copyright Ownership, and Publishing Ethics

Abstract

Background: Generative artificial intelligence (AI) tools, including Grabstract, NanoBanana, DALL-E, Midjourney, Adobe Firefly, and BioRender AI, have transformed scientific visualization by enabling the creation of publication-quality figures from text descriptions.

Aim: This study aimed to: (i) analyze current publisher policies on AI-generated visual content; (ii) evaluate copyright implications based on the latest legal guidance; (iii) examine practical implementation challenges, including detection technology limitations; and (iv) propose evidence-based declaration frameworks that balance transparency with applicability.

Methods: We analyzed editorial policies from major scientific publishers and examined copyright frameworks in major legal jurisdictions, specifically focusing on regulations from the US Copyright Office and the European Union.

Results: Most journals required AI disclosure for text generation but lacked specific guidance for visual content. The US Copyright Office (January 2025) and EU frameworks recognize copyright protection only when authors make substantial creative contributions through expressive inputs, modifications, or selection; pure AI outputs generated from prompts alone receive no protection. Furthermore, current publisher copyright transfer agreements inadequately address AI-generated content.

Conclusion: Transparent disclosure practices are essential to maintain research integrity. Journals must develop explicit policies addressing declaration requirements and copyright considerations. We propose a standardized "Artificial Intelligence Declaration" in methodology sections, detailing tool identification, generation scope, human contributions, and copyright verification, to address these gaps.