Journal policies on AI use have evolved rapidly since 2023. What was a gray area two years ago now has explicit policy at most major publishers. We surveyed 50 major journals across STEM, social sciences, and humanities to compile the most current picture of what's allowed.
As of 2026, virtually all major publishers (Elsevier, Springer, Nature, IEEE, ACM) require disclosure of AI tool use in manuscripts. The majority prohibit listing AI tools as co-authors. Most allow AI assistance for specific tasks while requiring disclosure. A small number of venues prohibit AI assistance entirely.
Grammar and spell checking, literature search assistance, code generation for analysis scripts, generating figures and charts from data, summarizing papers for background reading, and improving sentence clarity are broadly accepted when disclosed. These are considered tools, not content generation.
Using AI to draft sections of text, generate hypotheses, interpret results, or produce substantial portions of the paper typically requires explicit disclosure in your methods or acknowledgments section. The exact format varies by journal — check the submission guidelines for the specific disclosure language required.
Generating fabricated citations or references (a significant problem with earlier LLMs), listing AI as an author or contributor, using AI to generate data or experimental results, and using AI to bypass plagiarism checks are universally prohibited and constitute research misconduct.
Our recommendation: use AI tools freely for the tasks where they add clear value (grammar, search, code), disclose any use that materially affected the paper's intellectual content, never let AI generate citations without manual verification of every reference, and always check the specific policy of your target journal before submission.
Catch the issues in this article before a real reviewer does. Free for your first 2 papers.
Start Free Review →