AI-assisted coding tools can make software development faster and more accessible, but they also introduce new risks around code quality, security, copyright, attribution, transparency and research integrity. Researchers remain responsible for understanding, testing and documenting code produced with AI tools, particularly when it is shared publicly or used to support published research.

The guide is aimed at researchers and developers at all levels of coding experience. It explores the ethical and legal challenges associated with AI-assisted coding and provides practical advice for using these tools safely, responsibly and transparently.

Readers will learn how to review and test AI-generated code, disclose the use of AI tools, consider licensing and provenance, and improve the transparency, reproducibility and reusability of AI-assisted research software. The guide also discusses wider concerns, including data security, environmental sustainability and the potential deskilling of developers.

Its central principles are scrutiny, disclosure and accountability: AI-generated code should be carefully checked by humans, the use of AI should be clearly documented, and responsibility for the resulting software and research outputs must remain with the human authors.