Hallucinated citations: what journals and universities now require

Many publishers, conferences and universities now state who is responsible when an AI tool produces a wrong or non-existent reference. The answer is the same everywhere: the human author. Each entry below gives the organisation, a short quote, the link, and the policy date if one is shown.

Editorial bodies

ICMJE (Recommendations, updated January 2026). "Authors are responsible for citing references accurately and should be able to attest that the references cited support the associated statement." It also says "Referencing AI-generated material as the primary source is not acceptable." (ICMJE)
COPE (position "Authorship and AI tools", last reviewed 13 February 2023). "Authors are fully responsible for the content of their manuscript, even those parts produced by an AI tool." (COPE)

Publishers and journals

Springer Nature / Nature Portfolio (AI policy, no date shown). Lists "Fabricating data, citations or results" as not permitted. Authors "Remain fully accountable for originality, accuracy, and integrity." (Nature Portfolio)
Science journals (AAAS) (editorial policies, no date shown). "Authors are accountable for the accuracy of the work." They "must also ensure that all sources are appropriately cited." (Science)
Elsevier (generative AI policy for journals, updated June 2026). "Authors should always check sources, as AI-generated references and citations can be incorrect, hallucinated, or fabricated." Also: "Inclusion of fabricated references may lead to rejection of a manuscript." (Elsevier)
JAMA Network (JAMA Instructions for Authors, last updated 8 September 2026). "AI, LLM, and chatbots should not be used to generate or format references." Also: "Check all references for completeness, accuracy, and integrity." (JAMA)
APA Journals (policy on generative AI, created November 2023, last updated August 2025). It says "authors must verify any information and citations provided to them by an AI tool." (APA Journals)
APA Style (citing generative AI, 9 September 2025). It tells authors "you must check the output of the AI to ensure that the content is accurate to the best of your knowledge." (APA Style) An earlier APA Style post advised making "scrutiny of the primary sources a standard process" (created 7 April 2023, updated 15 September 2025). (APA Style, 2023)

Preprint servers and conferences

arXiv (moderation policy, no date shown). If AI tools produce "errors, mistakes, incorrect references, or misleading content" that authors include, "it is the responsibility of the author(s)." (arXiv)
arXiv, computer science section (post by section chair Thomas G. Dietterich, 14 May 2026). Citing arXiv's Code of Conduct, he wrote that "each author takes full responsibility for all its contents, irrespective of how the contents were generated." (Dietterich on X)
NeurIPS (2025 LLM policy for authors, no date shown). Authors must ensure "All content is correct (e.g., no citations of non-existent material)." Examples of violations include "using references generated by an LLM without conducting the due diligence to verify correctness, existence and appropriateness." (NeurIPS)
NeurIPS board (statement to Fortune, 21 January 2026). The board said "in 2025, reviewers were instructed to flag hallucinations." (Fortune) The statement responded to a report of hallucinated citations in accepted NeurIPS 2025 papers. A separate preprint analysed 100 such citations in 53 accepted papers (Ansari, 2026).
ICLR 2026 (blog, 19 November 2025). Hallucinated references "would be considered a Code of Ethics violation on the part of the paper's authors." (ICLR) The program chairs later wrote that "all papers with confirmed hallucinated references were desk rejected" (31 March 2026). (ICLR)
ICML 2026 (policy for LLM use in reviewing, no date shown). "Any hallucinated content in reviews is subject to disciplinary action according to our academic integrity policy." (ICML)

Universities

Penn State (Academic Integrity, "Limitations of AI Models", no date shown). "You need to carefully verify any citation and reference generated by an AI tool." (Penn State)
University of Edinburgh (guidelines for students, revision March 2026). Lists as misconduct: "Citing and referencing AI-found sources without reading and verifying them." (Edinburgh)
University of Greenwich (AI guidance for students, no date shown). Students should not "Treat AI-generated references as accurate without verifying their original sources." (Greenwich)
Stellenbosch University (memorandum, 21 May 2026). "Staff and students are encouraged to critically assess and verify and check all references generated or suggested by these tools." (Stellenbosch)

Courts

Courts have sanctioned lawyers for citing non-existent cases, for example the $5,000 penalty in Mata v. Avianca (S.D.N.Y., 22 June 2023) (sanctions order). Legal citations are outside the scope of this page.

Sources

All accessed 2026-10-24. Each entry gives the date shown on the source, if any.

icmje: ICMJE, Recommendations for the Conduct, Reporting, Editing, and Publication of Scholarly Work in Medical Journals; updated January 2026; sections II.A.4, IV.A.3.g.i and V.A. https://www.icmje.org/icmje-recommendations.pdf
cope: COPE position, "Authorship and AI tools"; last reviewed 13 February 2023; doi:10.24318/cCVRZBms. The live page blocked automated access; text checked in an Internet Archive copy of 12 October 2026. https://publicationethics.org/guidance/cope-position/authorship-and-ai-tools
nature: Nature Portfolio editorial policies, "Artificial Intelligence (AI)"; no date shown. https://www.nature.com/nature-portfolio/editorial-policies/ai
science: Science journals editorial policies, "Artificial intelligence (AI)"; no date shown. The live page blocked automated access; text checked in an Internet Archive copy of 14 June 2026. https://www.science.org/content/page/science-journals-editorial-policies
elsevier: Elsevier, "Generative AI policies for journals"; policy updated June 2026. https://www.elsevier.com/about/policies-and-standards/generative-ai-policies-for-journals
jama: JAMA, Instructions for Authors, "Use of AI in Publication and Research"; last updated 8 September 2026. The live page blocked automated access; text checked in an Internet Archive copy of 17 October 2026. https://jamanetwork.com/journals/jama/pages/instructions-for-authors
apa-journals: APA, "APA Journals policy on generative AI: Additional guidance"; created November 2023, last updated August 2025. Checked in an Internet Archive copy of 14 October 2026. https://www.apa.org/pubs/journals/resources/publishing-tips/policy-generative-ai
apa-style: McAdoo, Denneny & Lee, "Citing generative AI in APA Style: Part 1, Reference formats"; 9 September 2025. Checked in an Internet Archive copy of 25 September 2026. https://apastyle.apa.org/blog/cite-generative-ai-references
apa-chatgpt: McAdoo, "How to cite ChatGPT"; created 7 April 2023, last updated 15 September 2025 (kept for archival purposes). https://apastyle.apa.org/blog/how-to-cite-chatgpt
arxiv: arXiv moderation, "Policy for authors' use of generative AI language tools"; no date shown. https://info.arxiv.org/help/moderation/index.html
dietterich: Thomas G. Dietterich, post on X, 14 May 2026. https://x.com/tdietterich/status/2055000956144935055
neurips: NeurIPS 2025, "Policy on the Use of Large Language Models"; no date shown. https://neurips.cc/Conferences/2025/LLM
fortune: Sharon Goldman, Fortune, 21 January 2026 (quotes the NeurIPS board statement). https://fortune.com/2026/01/21/neurips-ai-conferences-research-papers-hallucinations/
ansari: Ansari, "Compound Deception in Elite Peer Review", arXiv preprint (not peer reviewed), 5 February 2026. https://arxiv.org/abs/2602.05930
iclr-nov: ICLR blog, "ICLR 2026 Response to LLM-Generated Papers and Reviews"; 19 November 2025. https://blog.iclr.cc/2025/11/19/iclr-2026-response-to-llm-generated-papers-and-reviews/
iclr-mar: ICLR blog, "A Retrospective on the ICLR 2026 Review Process"; 31 March 2026. https://blog.iclr.cc/2026/03/31/a-retrospective-on-the-iclr-2026-review-process/
icml: ICML 2026, "Policy for LLM use in reviewing"; no date shown. https://icml.cc/Conferences/2026/LLM-Policy
psu: Penn State Academic Integrity, "Limitations of AI Models"; no date shown. https://integrity.psu.edu/node/38
ed: University of Edinburgh, "Using generative AI in your studies: guidelines for students"; revision March 2026. https://information-services.ed.ac.uk/computing/comms-and-collab/elm/generative-ai-guidance-for-students/using-generative
gre: University of Greenwich, "AI guidance for students" (PDF); no date shown. https://www.gre.ac.uk/__data/assets/pdf_file/0022/364711/ai-guidance.pdf
su: Stellenbosch University, memorandum "Responsible Use of Generative AI in Referencing Practices"; 21 May 2026. https://files.su.ac.za/public/division-research-development/documents/2026-05/responsible-use-generative-ai-referencing-practices.pdf
mata: Mata v. Avianca, Inc., No. 1:22-cv-01461 (S.D.N.Y.), Opinion and Order on Sanctions, ECF 54, 22 June 2023. https://www.courtlistener.com/docket/63107798/mata-v-avianca-inc/