Academic journals’ AI policies fail to curb the surge in AI-assisted academic writing

Summary

OLDER ITEM (Feb 2026) included because the knowledge base has nothing on journal AI-disclosure policy effectiveness. Examines 5,114 journals and over 5.2 million papers: roughly 70% of journals have adopted AI policies, mostly disclosure requirements, yet detected AI-assisted writing rose sharply across disciplines with no significant difference between journals that have policies and those that do not. Full-text analysis of 164,000 publications finds only about 76 of the 75,000 post-2023 papers explicitly disclose AI use - roughly 0.1% - which makes declarative policy the wrong instrument and argues for verifiable, accountable frameworks instead.

Contribution

Queued — not yet read.

Method

Queued — not yet read.

Relevance to RISE

Surfaced by the 2026-09-07 literature sweep as a high-priority candidate for the RISE knowledge base (angle: research productivity of AI). Verified against: PNAS page returned HTTP 403 to WebFetch; verified instead via CrossRef https://api.crossref.org/works/10.1073/pnas.2526734123 (title, authors Yongyuan He and Yi Bu, PNAS, published 2026-02-24, journal-article) plus the abstract content (5,114 journals, 5.2M papers, 70% policy adoption, 164,000 full texts, ~0.1% disclosure). PMC mirror exists at PMC12956875. DOI and title absent from dedupe_titles.txt.

Critique / open questions

Queued — not yet assessed.

Key quotes

The rapid integration of generative AI into academic writing has prompted widespread policy responses from journals and publishers. (abstract)

However, the effectiveness of these policies remains unclear. (abstract)