Generative AI refers to the use of machine learning algorithms such as large language models to generate novel text or images based on training on vast datasets. The risks of biased or inaccurate training of such models and their tendencies to hallucinate plausible-sounding information such as made-up research paper titles have been written about many times over. Those who tread cautiously before embracing new technologies will take the time to educate themselves about their risks and limitations, and may even conclude that the risks are too great to adopt generative AI tools at this time.

However, despite their many problems, generative AI also has many benefits. For scientists who struggle with language or writing, generative AI may serve as an equalizer, allowing them to polish their manuscripts and improve their communications. Tedious tasks such as writing routine code can also be automated, freeing up time for more interesting scientific pursuits.

At Nature Methods, part of the broader Nature Portfolio, we realize that we can’t hold back the oncoming tide that is generative AI, but we can set some guidelines for its use in the publishing process. The key to responsible use, in our view, is transparency.

Nature Portfolio’s policies do allow authors to use generative AI tools to help write sections of their manuscript or edit their manuscripts for readability, but they must declare this use in the Methods section. We advise our authors to not take AI-generated text at face value, but carefully check and edit such text for accuracy. Note that a generative AI tool cannot be listed as an author, as, being non-human, it cannot take accountability for the content of a manuscript.

Nature Portfolio does draw the line at some uses of generative AI. We do not allow AI-generated artwork, including on our covers, mainly because of copyright issues and the legal murkiness around image creation that remains unresolved at this time. We do not use AI tools to write news pieces, editorials or other content, and expect that our freelancers and academic authors will also follow this rule. Some Nature Portfolio journals have explored using AI to write short pieces such as Research Highlights. However, the use of AI will always be declared, and such content is carefully edited and fact-checked by human editors.

It is also important that peer reviewers be aware of our AI policies. In particular, uploading manuscripts into generative AI tools is not allowed during the peer review process. Not only is this unfair to authors who trust that we will provide them with feedback from a human expert in their field, but feeding an unpublished manuscript to a large-language model also potentially compromises the confidentiality of the peer review process. Reviewers who use any AI tools to assist in writing a review should first ask the editor about whether their plans fall under an acceptable use case.

There is no doubt that Nature Portfolio’s AI policies will be continually reviewed and modified as the landscape of generative AI rapidly changes.