BY Rosemary
Wikipedia marked its 25th anniversary on Thursday by announcing a series of new commercial agreements with major artificial intelligence companies, a move that underscores how the online encyclopedia is adapting to a rapidly changing internet shaped by generative AI.
The Wikimedia Foundation, the nonprofit organization that operates Wikipedia, said it has entered into new business arrangements with AI developers including Amazon, Meta Platforms, Microsoft, Perplexity and France-based Mistral AI. The deals expand on earlier licensing agreements and are designed to formalize how large technology companies access and use Wikipedia’s vast store of human-edited knowledge.
Long regarded as one of the enduring pillars of the early internet, Wikipedia was built on the ideal of freely shared information created and maintained by volunteers around the world. That vision has increasingly been challenged by the rise of powerful technology platforms and AI systems that rely heavily on large-scale data scraping to train chatbots and other tools.
AI developers have drawn extensively from Wikipedia’s more than 65 million articles, raising questions about who bears the cost of supporting the infrastructure that makes such access possible. While Wikipedia’s content is free to read and reuse under open licenses, the servers and systems required to deliver that information at massive scale are not.
The Wikimedia Foundation said the new agreements will allow AI companies to access Wikipedia content “at a volume and speed designed specifically for their needs,” while helping the organization manage the growing technical burden caused by automated data collection. Financial terms were not disclosed.
The foundation began pursuing licensing arrangements with AI developers in recent years, signing Google as one of its first major customers in 2022 and later reaching agreements with smaller technology firms. The latest deals significantly broaden that effort at a time when AI-related traffic has surged.
Jimmy Wales, Wikipedia’s co-founder, said he supports the use of Wikipedia content in training artificial intelligence systems, particularly because the material is curated and moderated by humans. He has argued that models trained on high-quality reference material are more likely to produce reliable results than those drawing primarily from social media or other unfiltered sources.
At the same time, Wales has emphasized that cooperation with AI companies should come with financial responsibility. The Wikimedia Foundation has warned that automated scraping by bots—some of them designed to evade detection—has placed heavy strain on its servers. The organization reported last year that while human readership declined by about 8%, traffic from automated systems continued to rise sharply.
Those trends reflect broader changes in how people find information online, as AI-powered summaries and chat-based search tools increasingly provide answers directly rather than sending users to websites through traditional links.
Wikipedia remains the ninth most visited website in the world, supported by a global community of roughly 250,000 volunteer editors working across about 300 languages. Its operations are funded primarily by donations from millions of individuals, rather than advertising or subscriptions.
Wikimedia Foundation Chief Executive Maryana Iskander said maintaining the platform’s infrastructure carries significant costs, particularly as demand from technology companies grows. Iskander, who is stepping down later this month and will be succeeded by Bernadette Meehan, said donors contribute to support free access to knowledge—not to subsidize large commercial users.
Beyond licensing, the foundation is also exploring how artificial intelligence could be used internally to support Wikipedia’s volunteer editors. Wales said AI tools could eventually help with time-consuming maintenance tasks, such as identifying and replacing dead links or assisting editors in verifying sources, though he stressed that the technology is not yet capable of reliably writing full encyclopedia articles.
Artificial intelligence could also reshape how users interact with Wikipedia, he said, by allowing people to ask questions in natural language and receive answers drawn directly from cited passages within articles, rather than relying solely on keyword searches.
As Wikipedia reflects on its early days, Wales described the site’s founding era as one of experimentation and optimism, when contributors were motivated by the idea of building a shared public resource. He acknowledged, however, that online discourse has always had its darker sides, even before algorithm-driven platforms amplified conflict and division.
In recent years, Wikipedia has faced increasing political scrutiny, particularly from conservative figures who accuse the site of ideological bias. Some Republican lawmakers in the United States have launched inquiries into Wikipedia’s editing practices, arguing that they could influence public opinion and the AI systems trained on its content.
Technology entrepreneur Elon Musk has been among Wikipedia’s most vocal critics, accusing it of spreading propaganda and encouraging people to withhold donations. Musk last year launched an AI-driven project he has promoted as an alternative reference platform.
Wales dismissed such efforts as unlikely to rival Wikipedia’s depth and reliability, arguing that large language models tend to reproduce existing material rather than generate high-quality reference content on their own. He said AI-generated text often becomes less accurate and more incoherent when addressing obscure or complex topics.
Despite the criticism, Wales said he remains focused on keeping Wikipedia open, useful and sustainable as it enters its next quarter-century, navigating a digital landscape far different from the one in which it was born.
Original article
Comments