Jump to content

Artificial intelligence/Policies by project

From Meta, a Wikimedia project coordination wiki

This page documents AI policies and guidelines across Wikimedia projects. This list is incomplete. Please add any missing entries.

Content

[edit]

     Open proposals      Essays

AI content policies
Project Policy Status Date passed Description
Wikipedia editions
Azerbaijani Wikipedia (az) Wikipedia:Use of artificial intelligence

Vikipediya:Süni intellektdən istifadə[a]

Policy

Qayda

Proposal The use of LLMs is only accepted when it is of an auxiliary nature, such as improving grammar and style, summarizing sources, translating, or planning the structure of an article. In all of these cases, the output should be carefully checked. Unedited LLM-generated content is prohibited in mainspace.
Bulgarian Wikipedia (bg) Wikipedia:Use of artificial intelligence

Използване на изкуствен интелект[b]

Essay N/A TODO
Bengali Wikipedia (bn) Wikipedia:Large language models

বৃহৎ ভাষার মডেল[c]

Essay N/A TODO
Wikipedia:Criteria for speedy deletion § S15. Pages created entirely by artificial intelligence
উইকিপিডিয়া:দ্রুত অপসারণের জন্য বিচারধারা § স১৫. সম্পূর্ণরূপে কৃত্রিম বুদ্ধিমত্তা দ্বারা তৈরি পাতা[d]
Policy
নীতিমালা
26 August 2025 Pages created entirely by artificial intelligence
Catalan Wikipedia (ca) Wikipedia:Generating articles with artificial intelligence

Generació d'articles amb intel·ligència artificial[e]

Essay N/A TODO
Czech Wikipedia (cs) Wikipedia:Large language models

Wikipedie:Velké jazykové modely[c]

Guideline Doporučení Proposal TODO
German Wikipedia (de) Wikipedia:Artificial intelligence

Wikipedia:Künstliche Intelligenz[b]

Guideline

Richtlinie

15 February 2026 The use of LLMs to generate or edit content in mainspace is prohibited, besides translations, which should be fully checked for correctness.

The use of AI-powered tools in mainspace is only allowed if each suggested edit is verified by a human, and no AI-generated content is added beyond spelling and grammar corrections.

The use of LLMs in other namespaces is also prohibited.

Greek Wikipedia (el) Wikipedia:Artificial intelligence

Τεχνητή νοημοσύνη[b]

Essay N/A TODO
English Wikipedia (en) Wikipedia:Writing articles with large language models[a] Guideline 20 March 2026 The use of LLMs to generate or rewrite article content is prohibited, besides copyediting and translation.
Wikipedia:LLM-assisted translation[f] Guideline 10 February 2026 Text can be translated from a non-English Wikipedia to the English Wikipedia mainspace with an LLM-powered tool if:
  • The editor is fluent in both the source language and English;
  • All AI hallucinations and other core content policy violations have been checked for and removed;
  • All sources have been checked and are fairly reflected in the translation;
  • Each fact or claim likely to be challenged is cited to a reliable source;
  • Usual translation requirements are complied with.
Wikipedia:Speedy deletion § G15. LLM-generated pages without human review[d] Policy 4 August 2025 Pages exhibiting one or more of the following signs, which indicate that they could only plausibly have been LLM-generated and would have been removed by any reasonable human review, are eligible for speedy deletion:
  • Communication intended for the user
  • Implausible, non-existent, or nonsensical references
Spanish Wikipedia (es) Wikipedia:Writing articles with large language models

Wikipedia:Redacción de artículos con modelos de lenguaje grandes[a]

Policy

Política

29 January 2026 LLMs should not be used to create new Wikipedia articles from scratch, or to expand or replace existing entries.
Wikipedia:Criteria for speedy deletion § G12

Wikipedia:Criterios para el borrado rápido § G12[d]

Policy

Política

29 January 2026 Pages generated by LLM without human revision are eligible for speedy deletion.
Farsi Wikipedia (fa) Wikipedia:Creating an article with large language models ویکی‌پدیا:ساخت مقاله با مدل‌های زبانی بزرگ[a] TODO TODO TODO
Wikipedia:Translation with the help of large language models
ویکی‌پدیا:ترجمه با کمک مدل‌های زبانی بزرگ[f]
TODO TODO TODO
French Wikipedia (fr) Wikipedia:Generative artificial intelligence

Wikipédia:Intelligence artificielle générative[b]

Guideline

Recommandation

8 June 2025 Reasoned LLM use is tolerated for some tasks, but:
  • The generated content should conform to existing policies and guidelines, and be thoroughly verified;
  • The model and its use should be mentioned in the edit summary or on the talk page;
  • Familiarity with limitations of LLMs and the topic at hand is required;
  • The editor is responsible for their edits.
Finnish Wikipedia (fi) Help:Using AI

Ohje:Tekoälyn käyttö[c]

TODO Proposal TODO
Hebrew Wikipedia (he) Wikipedia:Artificial intelligence

ויקיפדיה:בינה מלאכותית[b]

TODO TODO TODO
Croatian Wikipedia (hr) Wikipedia:Artificial intelligence

Wikipedija:Umjetna inteligencija[b]

Essay N/A TODO
Hungarian Wikipedia (hu) Wikipedia:Artificial intelligence texts

Wikipédia:Mesterséges intelligenciával létrehozott szövegek[e]

TODO TODO There is consensus that such texts can be deleted based on our existing guidelines (discussion)
Indonesian Wikipedia (id) Wikipedia:Artificial intelligence

Wikipedia:Kecerdasan buatan[b]

TODO Proposal TODO
Italian Wikipedia (it) Wikipedia:Artificial intelligence

Wikipedia:Intelligenza artificiale[b]

Guideline

Linea guida

5 July 2025 The use of LLMs to generate original content is prohibited in both articles and drafts. LLMs can be used for stylistic review or paraphrasing, to suggest ideas, and to draft non-original content such as wikitext lists or tables in one's sandbox. LLM translations are treated the same way as other machine translations.

Users must be fully aware of the limitations of LLMs and properly assess their risks. Human review is necessary, and users take full responsibility for their edits. LLM use should be disclosed in the edit summary and on the talk page.

AI-powered semi-automated tools are prohibited.

Wikipedia:Speedy deletion § C3

Wikipedia:Cancellazioni immediate § C3[d]

Guideline

Linea guida

3 August 2025 Pages created through non-allowed use of AI are eligible for speedy deletion.
Japanese Wikipedia (ja) Wikipedia:Using large language models

Wikipedia:大規模言語モデルの利用[c]

TODO Proposal TODO
Korean Wikipedia (ko) Wikipedia:Speedy deletion criteria § G15. LLM-generated documents that have not been reviewed by humans
위키백과:빠른 삭제 기준 § G15. 사람의 검토를 거치지 않은 LLM 생성 문서[d]
Policy
정책
24 September 2025 This criterion applies to documents that were generated entirely by large language models (LLMs) and would have been modified after human review[3] to eliminate the problematic elements.
Dutch Wikipedia (nl) Wikipedia:AI and Wikipedia

Wikipedia:AI en Wikipedia[b]

TODO TODO TODO
Norwegian Wikipedia (no) Wikipedia:Guidelines for deletion § 23

Wikipedia:Retningslinjer for sletting § 23[g]

Guideline

Anbefaling

16 February 2026 Articles most likely produced with the help of LLMs or similar AI tools, and that haven't been significantly worked on or quality controlled language-wise and factwise, are eligible for speedy deletion.
Polish Wikipedia (pl) Wikipedia:Recommendations for the use of AI tools in Polish Wikipedia

Wikipedia:Zalecenia zastosowania narzędzi SI w polskiej Wikipedii[e]

TODO Proposal TODO
Portuguese Wikipedia (pt) Wikipedia:Large language models
pt:Wikipédia:Grandes modelos de linguagem[c]
Essay N/A Translation of May 2025 version of enwiki's Wikipedia:Large language models
Wikipedia:What Wikipedia is not § Wikipedia is not an inordinate repository of information

Wikipédia:O que a Wikipédia não é § A Wikipédia não é um repositório desordenado de informações[h]

Policy

Política

14 June 2025 LLM-generated content is considered pre-fabricated text and disallowed.
Romanian Wikipedia (ro) Wikipedia:Using artificial intelligence

Wikipedia:Folosirea inteligenței artificiale[b]

Essay N/A TODO
Wikipedia:Speedy deletion criteria § G13. Pages generated by language models, without human review.
Wikipedia:Criterii de ștergere rapidă § G13. Pagini generate de modele lingvistice, fără revizuire umană.[d]
Policy
Reguli
12 December 2025 It applies to any page that has one or more of the following signs indicating that it was almost certainly generated using a large language model (LLM), and that would have been removed at any reasonable human review (a careful reading of the text generated by the language model with verification of the existence of citations and their approximate accuracy) made prior to publication.
Russian Wikipedia (ru) Wikipedia:Neural network

Википедия:Нейросеть[a]

Guideline

Руководство

22 March 2026 In mainspace, unedited LLM content is prohibited. Content should be fully checked before being added, and disclosure in the edit summary or on the talk page is recommended. Content violating these rules should be deleted or corrected. In doubt, the content should be temporarily reverted while consensus is sought.

Bulk LLM-created articles should be deleted if they violate policies or guidelines. A lack of content-checking in several articles is enough to justify mass deletion.

Speedy deletion § C7
Википедия:Критерии быстрого удаления § C7[d]
Policy
Правила
23 August 2025 TODO
Serbo-Croatian Wikipedia (sh) Wikipedia:Speedy deletion § G15. Pages generated by artificial intelligence without human review Policy
Pravilo
4 May 2026 Translated from en:WP:G15.
Sinhalese Wikipedia (si) Large language models

විකිපීඩියා:Large language models[c]

Essay N/A Translated from the October 2024 version of enwiki's Wikipedia:Large language models.
Simple English Wikipedia (simple) Wikipedia:AI generated content[c] Guideline 16 July 2025 AI generated content must adhere to complexity guidelines, and not violate copyright laws. Editors are solely responsible for the content they post.
Wikipedia:Deletion policy § Articles[g] Policy 16 July 2025 Is clearly AI-generated. Pages very likely generated by a large language model (LLM) and recommended for deletion by at least two editors.
Slovene Wikipedia (sl) Wikipedia:Use of artificial intelligence

Wikipedija:Uporaba umetne inteligence[c]

Policy
Pravilo
30 January 2025 Use of LLMs is allowed for copyediting and other auxiliary tasks (such as translation), provided the user has prior experience performing these tasks without AI assistance. Any use must be explicitly disclosed, though it is generally discouraged. Newly generated LLM content is removed in accordance with local CSD guidelines.
Albanian Wikipedia (sq) Wikipedia:Large language models

Wikipedia:Modelet e mëdha gjuhësore[c]

Essay N/A Translation of some of a December 2025 version of enwiki's Wikipedia:Large language models
Wikipedia:Speedy tearing § General criteria
Wikipedia:Grisja e shpejtë § Kriteret e përgjithshme[d]
Policy
Politikat
Proposal AI-generated materials without any human scrutiny
Swedish Wikipedia (sv) Wikipedia:Artificial intelligence

Wikipedia:Artificiell intelligens[b]

Guideline

Riktlinje

Proposal An obviously AI-generated article can be deleted without needing to take into account any negative reactions from the author. In cases of doubt, the article can always be moved to a suitable sandbox.

AI tools can be helpful if used in a controlled manner, for example for technical and time-consuming tasks.

Silesian Wikipedia (szl) Wikipedia:Generative AI

Wikipedyjo:Generatywnŏ sztucznŏ inteligyncyjŏ[b]

policy 8 February 2025 It is not permitted to publish content generated by AI without rewriting it in own words. Listed typical AI slop errors are cause for long-term bans (for linguistic legacy protection reasons).
Telugu Wikipedia (te) Wikipedia:Large language templates

వికీపీడియా:పెద్ద భాషా నమూనాలు[c]

Essay N/A Translation of some of a November 2024 version of enwiki's Wikipedia:Large language models
Turkish Wikipedia (tr) Writing articles with large language models

Büyük dil modelleriyle madde yazma[a]

Essay

Deneme

N/A Translated from the November 2025 version of English Wikipedia's NEWLLM.
Wikipedia:Speedy deletion § G13. Pages created with artificial intelligence without human control can be deleted.
Vikipedi:Hızlı silme § G13. İnsan kontrolü olmaksızın yapay zekâ ile oluşturulan sayfalar silinebilir.[d]
Policy
Politika
22 November 2025 Pages created with artificial intelligence without human control can be deleted.
Ukrainian Wikipedia (uk) Написання статей з використанням LLM[a] TODO TODO TODO
Vietnamese Wikipedia (vi) Viết bài bằng mô hình ngôn ngữ lớn[a] TODO TODO Translated from the November 2025 version of English Wikipedia's NEWLLM.
Wikipedia:Large language models
Wikipedia:Mô hình ngôn ngữ lớn[c]
Essay N/A Translation of August 2025 version of enwiki's Wikipedia:Large language models
Chinese Wikipedia (zh) Wikipedia:Speedy deletion § G21

维基百科:快速删除 § G21[d]

Policy

方針

18 September 2025 Translated from en:WP:G15.
Wikipedia:Large language models

維基百科:大語言模型[c]

Essay

論述

N/A; failed related discussion April 2023 Largely translated from a past version of en:WP:LLMS. It goes over some issues of LLMs with regard to OR, V and NPOV, as well as copyright issues, but it does recommend using AI to paraphrase, summarise sources, or copyedit, contrary to the corresponding enwiki version. It notes the need to be familiar with the limitations of LLMs, and the need for editors to be responsible for their own edits and carefully check all of the output.
Wikipedia:Translating articles using large language models
维基百科:使用大型语言模型翻译条目[f]
TODO Proposal Translated from en:WP:LLMTRANSLATE.
Wikipedia:Writing articles with large language models
维基百科:使用大型语言模型撰写条目[a]
TODO Proposal Translated from en:WP:NOLLM.
Yue Chinese Wikipedia (zh-yue) Wikipedia:Content from computer generation

Wikipedia:電腦gen出嚟嘅內容[i]

Policy Proposal LLM-generated content contains issues with NPOV, V, OR, copyright and other policies. LLM content which contravenes policies will be considered disruptive editing. Editors are discouraged from using LLMs to assist editing in any way as the output is currently low quality; while banning generative AI technology is impossible due to the widespread use of it in commercial tools, editors must exercise caution and check carefully any output from a computer program. AI-generated pictures are not generally accepted for use on Wikipdia.
Wikibooks editions
English Wikibooks (en) Wikibooks:Artificial intelligence[b] Policy 14 December 2025 LLMs may not be used to generate or summarize material and ideas at Wikibooks, and their sources should not be blindly trusted. Translations made by LLMs are not allowed on Wikibooks.
Wikiversity editions
English Wikiversity (en) Wikiversity:Artificial intelligence[b] Policies 17 April 2026 AI may be used if disclosed in the edit summary with link to chatbot conversation and subject to user review, and a AI-generated notification template is applied to any page with substantial AI content.

Discussions

[edit]

     Open proposals      Essays

AI discussion policies
Project Policy Status Date passed Description
Wikipedia editions
Azerbaijani Wikipedia (az) Wikipedia:Use of artificial intelligence

Vikipediya:Süni intellektdən istifadə[a]

Policy

Qayda

Proposal LLM-generated content is prohibited in discussions. It can be accepted as a tool to help refine ideas in some cases, for example by correcting grammar or translating terms, although the generated content should be carefully checked.
German Wikipedia (de) Wikipedia:Artificial intelligence

Wikipedia:Künstliche Intelligenz[b]

Guideline

Richtlinie

15 February 2026 The use of LLMs outside of mainspace is prohibited.
English Wikipedia (en) Wikipedia:Talk page guidelines#LLM-generated Guideline Do not use LLMs to generate comments for discussions. (Grammar correction and similar refinements are acceptable.)
Italian Wikipedia (it) Wikipedia:Artificial intelligence

Wikipedia:Intelligenza artificiale[b]

Guideline

Linea guida

5 July 2025 LLMs should be avoided in discussions, although their use to reformulate one's comments is allowed. AI outputs may not be used as data or arguments in discussions.
Russian Wikipedia (ru) Wikipedia:Neural network

Википедия:Нейросеть[a]

Guideline

Руководство

22 March 2026 Unedited LLM content is prohibited in discussions, except as examples of LLM output. Edited LLM content is only allowed if labeled as such, and can otherwise be closed, collapsed or removed.
Chinese Wikipedia (zh) Wikipedia:Talk page guidelines § How to provide feedback

Wikipedia:討論頁指引 § 如何提供意見[j]

Guideline

指引

30 November 2025 LLM-generated comments should be manually proofread. Clearly LLM-generated comments may be collapsed or removed.

Images and media

[edit]

     Open proposals

AI media policies
Project Policy Type Date passed Description
Multilingual projects
Wikimedia Commons AI-generated media[b] Guideline 12 June 2024 Per the Commons project scope, only media that are realistically useful for an educational purpose should be hosted on Commons. Just because an AI image is interesting, pretty, or looks like a work of art, that doesn't mean that it is necessarily within the scope of Commons. While some AI-generated media fall within our scope, media that lack a realistic educational use may be nominated for deletion.
AI images of identifiable people 15 February 2026 Images of identifiable people created by AI are not allowed on Commons unless the image in question was published by the person it depicts, or the image in question is the subject of non-trivial coverage by reliable sources. Photographs of people that are subsequently altered by AI are not allowed on Commons unless the original, unaltered image is linked and the AI-modified file is tagged.
Wikipedia editions
Azerbaijani Wikipedia (az) Wikipedia:Use of artificial intelligence

Vikipediya:Süni intellektdən istifadə[a]

Policy

Qayda

Proposal The use of AI images is prohibited, except for images used by reliable sources and disclosed by those sources as AI. In that case, the use must still be appropriate and non-misleading, and the image should be disclosed as AI in its caption.
German Wikipedia (de) Wikipedia:Artificial intelligence

Wikipedia:Künstliche Intelligenz[b]

Guideline

Richtlinie

15 February 2026 As an exception to the AI guideline, LLM-generated or -modified images may be used sparingly. Outside of the AI topic area, such images must only be used with prior talk page consensus. AI-generated and -modified images must always be declared as such and cannot be used to replace unavailable real images, such as a person's portrait.
English Wikipedia (en) AI-generated images[k] Policy 14 April 2025 Most images wholly generated by AI should not be used in mainspace, especially for named people and in technical or scientific subjects such as anatomy and chemistry. This rule should be applied "subject to common sense and with a number of exceptions". Obvious exceptions include articles about AI and articles about notable AI-generated images; other categories of exceptions may arise through further community discussion. Community members have largely rejected making exceptions merely because an image lacks obvious errors, or because no free non-AI-generated images are available.
Italian Wikipedia (it) Wikipedia:Artificial intelligence

Wikipedia:Intelligenza artificiale[b]

Guideline

Linea guida

5 July 2025 AI-generated media should not be used, except to illustrate topics related to AI image generation. They should never depict living persons, and should be indicated as AI-generated through templates and captions.
Chinese Wikipedia (zh) none As of 5 April 2026, the Chinese Wikipedia appears to have no restrictions on the use of AI-generated or -modified images.
Wikibooks editions
English Wikibooks (en) Wikibooks:Artificial intelligence[b] Policy 14 December 2025 Any permissible media made with the help of an LLM must be explicitly marked as such in both the edit summary and the page's discussion page, and include the date of generation/addition, the tool and version, and the prompt(s). If incorporated into a page multiple times, each instance must be documented separately, including on the talk page.

WikiProjects

[edit]

     Inactive/Closed

AI-related WikiProjects
Project WikiProject Date founded Goals
Multilingual projects
Wikidata Embedding Project[l] 21 November 2024[note 1] Enhance the search functionality of Wikidata by integrating vector-based semantic search
Artificial Intelligence[m] 21 March 2023 Track and organize AI-related content; track applications and discussions in the interface of AI with Wikidata
Wikipedia editions
German Wikipedia (de) AI and Wikipedia

KI und Wikipedia[n]

1 July 2024 TODO
English Wikipedia (en) AI Cleanup[n] 4 December 2023 Scrutinize AI-generated content
AI Tools 17 November 2025 Develop AI-powered tools for purposes other than content generation
Finnish Wikipedia (fi) AI Control Room

Tekoälyvalvomo[n]

9 November 2025 TODO
French Wikipedia (fr) AI Observatory

Observatoire des IA[n]

7 October 2024 TODO
Japanese Wikipedia (ja) AI記事除去[n] 2 February 2026 TODO
  1. Officially launched on 1 October 2025

Wikidata items in use

[edit]