The Wikimedia Foundation today said Wikipedia and Wikidata have been added to the DPGs for AI Collection, a curated set of six digital public goods that the Digital Public Goods Alliance and three partner organisations put forward as material for responsible and equitable AI development. The news arrived as an update dated 19 November 2026 at the top of a Foundation post first published on February 12, 2025, when Wikipedia itself was recognised as a digital public good.

In Short

Wikipedia and its sister database Wikidata are now part of a hand-picked group of six free resources that a UN-endorsed alliance points to for building AI in the public interest. Both already sit underneath much of what chatbots and AI search tools tell people, so the label puts an official stamp on a role that also brings heavy automated traffic to Wikimedia's servers. Nothing in the update changes how the content can be used or accessed, and the Foundation does not name the other four members or the criteria they met.

A shortlist built for the public interest

The update is brief, at three paragraphs. According to the Wikimedia Foundation, the DPGs for AI Collection - which it also calls the DPG4AI Collection - is "a curated set of six digital public goods (DPGs) that can be used for responsible and equitable AI development," described as "tools that place the public interest at their core."

The intended users are named. Together with the other members of the collection, Wikipedia and Wikidata "provide open tools and knowledge that local developers, civil society, multilaterals, and governments can use as building blocks to develop and adapt AI solutions to their specific cultural and socioeconomic needs," the Foundation wrote. The reasoning is aimed at the gap between global AI products and local conditions. "This is particularly important in low-resource contexts and underrepresented domains, since ready-made AI tools can be ill-suited for what communities require and they may reinforce existing biases or inequalities."

Entry is not automatic. Wikimedia says the collection "has strict criteria" and is co-stewarded by the United Nations International Computing Centre (UNICC), the Open Data Institute (ODI), the Digital Futures Lab and the DPGA Secretariat. The Foundation said it "appreciates the recognition and inclusion of two of our biggest open source knowledge projects in this collection for advancing trustworthy, accessible, and equitable AI solutions for the public interest."

One line points beyond content. "Wikimedia's contribution to responsible AI goes beyond providing freely licensed content and data," the update states, adding that the Foundation "is also investing in infrastructure that gives communities greater control over how AI tools are developed and used." What that infrastructure is, how much is being spent, and when it arrives are not stated.

Item Detail
Collection name DPGs for AI Collection (DPG4AI); the DPGA's own page is titled "DPGs for Responsible AI (DPG4AI)"
Size Six digital public goods, according to Wikimedia
Wikimedia projects included Wikipedia, Wikidata
Co-stewards UNICC, Open Data Institute, Digital Futures Lab, DPGA Secretariat
Intended users Local developers, civil society, multilaterals, governments
Stated focus Low-resource contexts and underrepresented domains
Other four members Not named in Wikimedia's update
Date announced by Wikimedia November 19, 2026

Three recognitions since February 2025

The post has grown by accretion. Its newest update sits at the top, an October 2025 update below it, and the original February 2025 announcement underneath, while the byline now carries the November 19, 2026 date. Read from the bottom up, it records three separate steps.

Date Project What happened Wikimedia's wording
February 12, 2025 Wikipedia Recognised as a digital public good by the DPGA "officially been recognized as a digital public good"
September 2025 (an update dated October 29, 2025 says "last month") Wikidata Recognised as a digital public good, the second Wikimedia project to be listed "the second Wikimedia project to be officially recognized"
November 19, 2026 Wikipedia and Wikidata Included in the DPGs for AI Collection "now a part of the DPGs for AI (DPG4AI) Collection"

The distinction matters. The 2025 decisions placed both projects in the DPGA's general registry, which according to the Foundation covers "open source-software, data, AI models, standards, and content created for the public interest." Today's step is narrower: selection from that registry into a small set meant specifically for AI work.

What Wikidata brings

Wikidata is the less familiar of the two names outside technical circles. According to the October 2025 update, it is "a free and open knowledge base, which can be read and edited by both people and machines, and acts as central storage for the structured data (including text, images, and multimedia) of the other Wikimedia projects." That structured data, the Foundation wrote, "is also available for use under a free license to other external applications and services."

Wikimedia lists institutions it works with through Wikidata: the United Nations Educational, Scientific and Cultural Organization (UNESCO), Europeana, the Library of Congress and OpenStreetMap. The Foundation presents Wikidata as a case of open data serving the UN Sustainable Development Goals, writing that it shows how such data "can help to advance education, gender equality, innovation, economic growth, and public institutions while reducing inequalities."

Machine readability is the reason Wikidata belongs in an AI collection. A knowledge base built so that software can query it directly is useful to anyone assembling or checking a model's facts - and, as Wikimedia's own infrastructure reports show, it is also exposed to heavy automated demand.

The 2025 case for Wikipedia

The original announcement explains what the alliance is. The DPGA, according to the Foundation, is "a multistakeholder initiative that maintains a Registry of Digital Public Goods" and is endorsed by the UN Secretary-General "in support of open source technologies that contribute to the advancement of the Sustainable Development Goals." In practice, that means open source software, data, models, standards and content that clear the alliance's standard.

Liv Marte Nordhaug, chief executive of the DPGA Secretariat, was quoted at the time: "Wikipedia's recognition as a digital public good is a testament to the power and importance of open access to knowledge. Wikipedia stands as a prime example of how technologies can drive equitable and unrestricted access to information, accelerating the attainment of the Sustainable Development Goals worldwide."

Wikimedia describes Wikipedia as "the world's largest online encyclopedia and only top-visited website operated by a non-profit organization," maintained by "a global community of nearly 260,000 volunteers in over 300 languages." That volunteer figure dates from the February 2025 text. When Wikipedia turned 25, PPC Land reported figures of nearly 250,000 volunteer editors, 65 million articles in more than 300 languages and nearly 15 billion views a month.

Rebecca MacKinnon, Vice President of Global Advocacy at the Wikimedia Foundation, framed the 2025 recognition in policy terms. "The Wikimedia Foundation works with affiliate organizations and volunteer Wikipedians across the world to advocate for policies that protect and support Wikipedia and other digital public goods upon which the free knowledge ecosystem depends," she said. "We look forward to working with the Digital Public Goods Alliance, along with other organizations and communities that create and maintain digital public goods, to build a better internet that serves the public interest."

The UN agenda behind the label

The recognitions sit inside a longer advocacy campaign. In 2024, according to the Foundation, its staff and Wikipedia volunteers took part in the UN General Assembly's Summit of the Future and in drafting the Global Digital Compact, which Wikimedia calls "the UN's blueprint for global governance of digital technology and artificial intelligence."

In an open letter in early 2024, the Foundation and Wikimedia affiliates called on UN Member States to:

  • "Protect and empower communities to govern online public interest projects."
  • "Promote and protect digital public goods by supporting a robust digital commons from which everyone, everywhere, can benefit."
  • "Build and deploy Artificial Intelligence (AI) and machine learning (ML) to support and empower, not replace, people who create content and make decisions in the public interest."

The third demand reads differently now that Wikipedia and Wikidata sit in an AI collection. "Support and empower, not replace" is a condition, and the DPG4AI listing is about making the two projects easier to build on. According to the Foundation, the 2025 recognition "strengthens these advocacy efforts," and it will "continue working with the UN and other international institutions, governments, and civil society partners to ensure that digital public goods like Wikipedia are protected, supported, and accessible to all."

How a project reaches the collection, according to the DPGA

Wikimedia's update links to the alliance's collection page and to the criteria document, but does not describe the process. The DPGA's pages set out the path in four stages.

  1. Registry first. A solution has to be a digital public good before it can be considered. The collection page tells projects that are not yet DPGs to start by applying to become one.
  2. Annual review. According to the DPGA's registry, DPG status is valid for one year from approval, after which the project must be reassessed.
  3. AI-specific criteria. The DPGA Secretariat developed a set of criteria built on the DPG Standard, in collaboration with UNICC, ODI and Digital Futures Lab and based on community feedback. The criteria are published in the alliance's dpg-standard repository on GitHub.
  4. Rolling additions. Existing DPGs can ask to be considered by contacting the alliance, and the collection is updated on a rolling basis.

Why the listing lands differently in marketing

For advertisers, publishers and search marketers, Wikipedia is not an abstract commons. It is one of the most important inputs to the systems that now answer questions before anyone clicks a link.

The reference layer under AI answers

The dependence is old and well documented. In October 2024, PPC Land reported that Wikipedia is almost always the largest source of training data for large language models, alongside Kevin Indig's August 2024 estimate that 84% of Wikipedia's traffic came from Google organic search. The relationship runs in both directions: Google sends readers to Wikipedia, and Google's AI products lean on it. When SISTRIX examined 10,000 keywords in Google's US AI Mode, Wikipedia was the most cited domain with 12.02% of appearances, ahead of Google.com at 7.41%.

Money has followed. By Wikipedia's 25th anniversary in January 2026, Ecosia, Microsoft, Mistral AI, Perplexity, Pleias and ProRata had joined Amazon, Google and Meta as Wikimedia Enterprise partners, paying for high-volume access to the same content anyone can read for free.

There have been challengers. xAI launched Grokipedia with 885,279 articles in October 2025 as an alternative, and by August 2026 PPC Land reported that Grokipedia had made no article text changes after April 24, leaving 13,002 human suggestions unresolved.

For companies, the practical effect is indirect but real. What Wikipedia articles and Wikidata entries say about a brand, a product or an executive already feeds the assistants and AI search features that marketers track. Today's listing does not change that mechanism. It formally commends it to developers and governments building their own systems, with a stated focus on low-resource contexts.

The cost of being everyone's raw material

Endorsement and strain arrive together. On October 5, Wikimedia said agents it believes OpenAI operates may have contributed to a partial disruption of the Wikidata Query Service in May. According to that account, the agents ran hundreds of thousands of data queries against the service and crawled millions of pages, mainly on Wikidata and Wikimedia Commons. Bots made up 65% of the most resource-consuming traffic on Wikimedia's projects, and bandwidth had risen 50% since 2024 because of bot activity, PPC Land reported. Large-scale web scraping is not free for the site being scraped, even when the content is.

Wikimedia's legal position draws a related line. On October 29, the Foundation asked a federal judge in the New York Times v. OpenAI litigation to treat AI training and AI retrieval differently, arguing that training is likely fair use but that retrieval which ignores Wikipedia's licence terms - attribution and ShareAlike under CC BY-SA 4.0 - likely is not. The Foundation also said scraping should follow its public API rate limits or go through its Enterprise offerings.

None of that appears in today's update. The DPG4AI text speaks of building blocks, local adaptation and community control, and says nothing about attribution, rate limits or paid access. As a nonprofit, Wikimedia says its projects are "powered primarily through donations." The listing adds a public-interest credential; whether it brings any support for the servers that the new users will query is not addressed. Who pays for that capacity remains, on today's evidence, mostly donors and the companies paying for Enterprise access.

What the update does not say

The update names neither the other four members of the collection nor the criteria Wikipedia and Wikidata met, and gives no date for the alliance's decision. It does not describe the infrastructure Wikimedia says it is investing in, carries no comment from the DPGA or the co-stewards, and does not say whether the listing affects licensing, Wikimedia Enterprise terms or API limits.

Key facts

  • Announcement: Wikipedia and Wikidata are part of the DPGs for AI (DPG4AI) Collection, announced by the Wikimedia Foundation on November 19, 2026
  • Collection: six digital public goods for "responsible and equitable AI development," according to Wikimedia; the DPGA titles its page "DPGs for Responsible AI (DPG4AI)"
  • Co-stewards: United Nations International Computing Centre (UNICC), Open Data Institute (ODI), Digital Futures Lab, DPGA Secretariat
  • Earlier recognitions: Wikipedia as a digital public good, February 12, 2025; Wikidata, September 2025 (per an update dated October 29, 2025)
  • Source document: "Wikipedia Recognized as a Digital Public Good," Wikimedia Foundation, first published February 12, 2025, updated October 29, 2025 and November 19, 2026: https://wikimediafoundation.org/news/2026/11/19/wikipedia-recognized-as-a-digital-public-good/
  • Collection page: https://www.digitalpublicgoods.net/dpgs-for-ai
  • Criteria: DPGA dpg-standard repository, branch collection-dpg-for-ai: https://github.com/DPGAlliance/dpg-standard/tree/collection-dpg-for-ai/collections/dpgs-for-ai
  • People quoted (2025): Liv Marte Nordhaug, CEO, DPGA Secretariat; Rebecca MacKinnon, Vice President of Global Advocacy, Wikimedia Foundation
  • Organisation: the Wikimedia Foundation is a United States 501(c)(3) tax-exempt organisation based in San Francisco

Timeline

Summary

Who: The Wikimedia Foundation, which operates Wikipedia and Wikidata, and the Digital Public Goods Alliance, which co-stewards the DPGs for AI Collection with the United Nations International Computing Centre, the Open Data Institute and the Digital Futures Lab.

What: Wikipedia and Wikidata are now part of the DPGs for AI Collection, a curated set of six digital public goods that local developers, civil society, multilaterals and governments can use as building blocks for AI. Both projects were already registered digital public goods, Wikipedia since February 2025 and Wikidata since September 2025.

When: Wikimedia announced the inclusion today, November 19, 2026, in an update to its February 12, 2025 post.

Where: The collection is global, with a stated focus on low-resource contexts and underrepresented domains where ready-made AI tools may not fit local needs.

Why: According to Wikimedia, open tools and knowledge help communities adapt AI to their cultural and socioeconomic needs without reinforcing existing biases. For marketers and publishers, the listing formalises the role Wikipedia and Wikidata already play as reference material for AI systems, at a time when Wikimedia says bot traffic is straining its infrastructure and argues in court that AI retrieval should credit its sources.