Springer Nature launches framework on use of artificial intelligence

Springer Nature has updated its editorial policies on the use of artificial intelligence, introducing a framework designed to help researchers, peer reviewers and editors assess when and how AI tools can be used responsibly throughout the research publishing lifecycle.

The publisher says the changes respond to growing demand for clearer and more practical guidance as AI becomes increasingly embedded in research and publishing workflows. More than 70% of researchers see publisher guidance on AI as increasingly important, according to Springer Nature.

The news comes on the same day that Oxford University Press launched its own set of guidelines around AI use by authors.

The updated policies build on the publisher’s existing AI principles of accountability, transparency and human oversight, as well as confidentiality in editorial assessment.

Rather than focusing simply on whether AI has been used, the new framework asks researchers, reviewers and editors to consider how the technology has been used, its potential impact, the level of risk involved and whether appropriate human oversight has been maintained.

Springer Nature says the framework is intended to distinguish between relatively low-risk applications of AI and uses that could have a more significant influence on research or publishing decisions.

Lower-risk applications could include improving language, cleaning data or helping to structure comments. Uses that could influence interpretation, evaluation or decision-making – including suggesting research methods, identifying patterns in data or helping to interpret results – require greater assessment, transparency and human oversight.

Some applications remain prohibited. Springer Nature says AI must not replace accountable human judgement or compromise the integrity of the publishing process. This includes using AI to generate research conclusions or delegating peer review to an AI tool.

The updated policies also place an emphasis on disclosure. Springer Nature says appropriate declarations of AI use will enable greater transparency and informed assessment, while recognising that responsible AI-assisted activities are becoming routine elements of scholarly research and communication.

Harsh Jegadeesan, Chief Publishing Officer at Springer Nature, commented: “We believe AI can accelerate discovery and improve research communication when used responsibly. Our updated AI policy guidance makes clear that human accountability remains the cornerstone of research and publishing and helps researchers use AI responsibly, transparently and with confidence.”

Ellie Gendle, Head of Journals Policy at Springer Nature, added: “AI use in research and research communications is evolving continuously, and as it continues to become more embedded it is increasingly important to be able to confidently distinguish between different use cases. Policies and guidance that apply an assessment and management approach to AI use can help academics to navigate this evolution responsibly, transparently and with greater confidence, whilst ensuring clear protections where greater risks to integrity or accountability may arise.

“Our new guidance continues to make clear that AI is not a substitute for human judgement and accountability, whilst recognising that, assuming it is done with care, responsibility and appropriate disclosures, AI can and will be applied in an increasing range of research and publishing activities.”