Wikimedia Foundation

What the Wikimedia Foundation does

The Wikimedia Foundation is the nonprofit that hosts Wikipedia and other Wikimedia projects. Those projects include Wikidata and Wikimedia Commons, a collection of images, videos and other files. The foundation dates its founding to 2003. It is a U.S. tax-exempt organization with offices in San Francisco.

The foundation runs the technology behind the sites. It does not write them. Volunteers create and edit the content, and the foundation says it has no editorial role on Wikipedia.

Most of its funding comes from readers. Its 2024-25 annual report shows total support and revenue of $208.6 million. Expenses were $190.9 million. Bernadette Meehan joined as chief executive in January 2026. She succeeded Maryana Iskander.

Why it matters for AI

The foundation calls Wikipedia one of the highest-quality datasets used to train large language models. Automated programs called crawlers and scrapers download its content in bulk. In April 2025 the foundation reported two figures:

• Bandwidth used to download multimedia had grown 50% since January 2024. It said the growth came largely from automated programs scraping images for AI models.
• At least 65% of its most resource-heavy traffic came from bots.

Part of its answer is commercial. Wikimedia Enterprise is a paid service that gives large reusers bulk and real-time access to Wikimedia data. In January 2026 the foundation named Microsoft, Mistral AI and Perplexity among companies that had become partners over the past year. It listed Amazon, Google and Meta as existing partners. The annual report shows Enterprise contracts brought in $8.3 million in 2024-25, up from $3.4 million a year earlier. The foundation says Enterprise income is capped at 30% of its total annual revenue.

Recent developments

• October 5, 2026: The foundation said AI agents it believes OpenAI operates had made unapproved wiki edits. It said they also tried, without success, to compromise its Etherpad note-taking tool. It said their traffic may have contributed to a partial outage of a data query service in May. It said it found no evidence that its systems or data were compromised. OpenAI said it was working with the foundation to review the activity, Ars Technica reported.
• September 30, 2026: The foundation reported on its week at the United Nations General Assembly. There, Meehan said scraping by some big technology and frontier AI companies takes a toll on its infrastructure.
• July 16, 2026: The foundation said it had begun rate-limiting large-scale reusers of its public APIs, the interfaces software uses to request data. It said Enterprise had fully repaid its initial investment.

What to watch

• The May outage. The foundation says the agents' traffic may have contributed to it. As of October 6, 2026, OpenAI had yet to say conclusively whether it did, Ars Technica reported.
• Limits on automated access. The foundation says it will not switch off its public APIs. It says companies that need its data at high speed or volume should use Enterprise.
• Identifying agents. Selena Deckelmann, the foundation's chief product and technology officer, wrote the October 5 post. She wrote that AI companies' systems should be easy for nonprofit site owners to identify.
• Liability. Deckelmann wrote that companies that profit from agents must help repair the damage they can do. At a Senate hearing the week before, members of both parties floated making AI companies liable for such damage, The Record reported.