Wikipedia urges AI firms to use its paid API, stop scraping
The Wikimedia Foundation has asked AI companies to credit Wikipedia and obtain data via its paid Enterprise API, citing bot-driven strain and an 8% drop in human page views.
Wikipedia’s parent, the Wikimedia Foundation, has called on AI developers to access its content through the paid Wikimedia Enterprise API and to provide clear attribution, arguing that large‑scale scraping has strained its infrastructure while reducing referrals that sustain the volunteer‑run encyclopaedia.
In its report, the foundation has asked AI companies to do two things: credit Wikipedia as a source and financially support the service by using Wikimedia Enterprise, an opt‑in, paid product designed for high‑volume reuse.
The foundation said this route enables accurate, timely data delivery without “severely taxing” Wikipedia’s servers and directly supports its non‑profit mission.
Wikimedia Enterprise offers enterprise‑grade APIs, contracts and SLAs, plus metadata and “credibility signals” meant to help downstream systems detect vandalism and interpret content at scale. The service is positioned specifically for AI, search and knowledge‑graph applications.
Traffic has fallen as bots evade detection
The foundation reported that traffic once thought to be human is in fact sophisticated bots. After updating its detection systems and reclassifying data from March–August 2025, it found that unusually high visits in May and June were driven by bots built to evade detection. Once reclassified, human page views were down roughly 8% year over year.
The new appeal to AI reusers followed those findings, with the foundation warning that fewer visits can mean fewer volunteer contributions and fewer donations — a feedback loop it wants platforms to counter through attribution and paid access.
Wikimedia Foundation CEO Maryana Iskander — who has led the organisation since 2022 and has planned to hand over the role by January 2026 — has previously argued that companies relying on Wikipedia should help fund the public good it provides. The call for responsible reuse has arrived amid that orderly transition.
Wikipedia’s model depends on transparent sourcing and community participation. The foundation has said that clear attribution and enterprise access can preserve that cycle — ensuring people can verify information at the source and that the infrastructure hosting free knowledge remains sustainable in the generative‑AI era.


