Effect of generative artificial intelligence on academic publishing ethics: The role of user agreements

Abstract

Generative artificial intelligence (AI) such as ChatGPT are rapidly expanding in scientific research and publication due to their excellent language understanding, generation and knowledge reasoning capabilities, but the accompanying ethical issues of publication cannot be ignored. User agreements can provide behavioural guidance for users to use generative AI tools, but its ethical guiding role in academic publishing is unclear. This study aims to investigate the ethical implications of generative AI in academic publishing, with a specific focus on the role of user agreements. We selected 98 large language models (LLMs) which were divided into 23 series, and analysed their user agreements and privacy policy content to explore their ethical guiding role in academic publishing. The survey study found that LLMs user agreements had relevant provisions on intellectual property, data privacy and security, quality and integrity, and so on, but there were still controversial and needed to be further improved contents, such as the ownership of LLMs output, and continuous supervision and update of user agreements. The publishing community should collaborate closely with academics, technical developers, and legal experts to promote the formation of an ethical framework for the application of generative AI technology in academic publishing. This study systematically analyses the guidance of user agreements for LLMs in academic publishing ethics, filling a gap in the literature on ethical norms in the field of academic publishing for generative AI.