Oxford University Press (OUP) has updated its author and editor guidelines on the responsible, appropriate, and transparent use of artificial intelligence. The new policy outlines what forms of AI use should be disclosed when preparing manuscripts for its journals.
A recent global survey of more than 2,600 researchers revealed that nearly three‑quarters (74%) were uncertain about what AI use should be disclosed, despite strong support for disclosure. The survey also found that 44% of researchers were concerned disclosure might negatively affect perceptions of their work, while only 36% recorded their AI use during research.
The updated guidelines aim to support researchers in recording and disclosing AI use with confidence. They emphasize disclosure across both books and journal articles, human oversight and accountability, and protection of intellectual property and confidentiality when using AI tools with unpublished work, copyrighted material, or sensitive content.
Survey findings indicated that 64% of researchers reported AI as beneficial to their work, with common uses including discovering existing research (55%), summarizing research (46%), and editing manuscripts (45%). However, trust in AI outputs remains low, with only 20% of respondents expressing confidence in them. Many researchers cited concerns about accuracy, bias, and reliability, though 83% reported established practices for checking AI‑generated citations and 80% cross‑checked outputs for reliability.
Attitudes toward AI varied across disciplines and career stages. Profiles ranged from “persistent skeptics,” most common in the arts and humanities, to “carefree casuals,” more prevalent among later‑career researchers. “Frustrated trailblazers” were identified as benefiting most from AI but also expressed concern about the administrative burden of disclosure.
The survey highlighted inconsistent disclosure practices. Nearly half of journal authors and book authors admitted they had not fully disclosed AI use in manuscript preparation, while 25% of peer reviewers reported incomplete disclosure, particularly in language editing. Respondents emphasized the need for clearer, coherent, and reasonable rules, with some suggesting that minor uses of AI should not require disclosure.
Looking ahead, many researchers expressed a desire for standardized and mandatory AI disclosure policies, while others advocated for normalization of AI as a tool with minimal reporting requirements. OUP stated that its updated guidelines are intended to provide clarity and confidence for authors and editors as global efforts continue to establish a reporting standard for AI disclosure in research.
Click here to read the original press release.
