Wikipedia seeks compensation from AI companies using its data

Wikipedia is working with big tech on deals similar to its arrangement with Google, the online encyclopedia's co-founder, Jimmy Wales, said on Wednesday, in a bid to help the firm monetize AI companies' heavy reliance on its content.

Speaking in an interview at the Reuters NEXT summit in New York, Wales said that tech companies' usage of freely available Wikipedia knowledge to train their large language models results in cost surges that Wikipedia's nonprofit operator must bear.

"The AI bots that are crawling Wikipedia are going across the entirety of the site ... So we have to have more servers, we have to have more RAM and memory for caching that, and that costs us a disproportionate amount," Wales said.

While the content of Wikipedia remains free for individuals under its license, the high-volume, automated access by for-profit entities is a different matter, Wales said. He noted that a deal has already been signed with Alphabet's Google and that discussions with other firms are ongoing.

The Wikimedia Foundation, which operates Wikipedia, struck a deal with Google in 2022 to have the tech giant pay for training access to Wikipedia content, which is a crucial part of data that companies like OpenAI and Meta Platforms use to train their AI models.

The foundation's primary source of income is small donations from the public, which Wales said are not intended to underwrite the development of multibillion-dollar commercial AI products.

Just today, Wikipedia is launching its annual donation campaign, in which the Wikimedia Foundation asks its readers to contribute financially.

“Most people donate because Wikipedia is the internet we were promised: useful, free to use, and filled with reliable, human-created knowledge. If you agree, consider giving $25 or even just $2.75,” the message at the beginning of Wikipedia articles states.

While the company does not state that this year’s fundraiser is extraordinary or in any way related to the additional costs stemming from AI, it comes amid ongoing discussions about fairly compensating Wikipedia for how intensively AI companies rely on its data.

"Wikipedia is supported by volunteers. Those people are donating money to support Wikipedia, and not to subsidize OpenAI, costing us a ton of money. That doesn't feel fair," said Wales.

The push for more licensing places the world's largest repository of free knowledge in a potential standoff with the burgeoning AI industry. It raises fundamental questions about who should bear the cost for the vast datasets that fuel the AI revolution and whether for-profit companies have an obligation to compensate the public and nonprofit sources that help build their technology.

The conversation on this topic is live. Join in the discussion.

Asked if Wikipedia would take legal action against AI companies using its content without paying for training access, Wales said: "I don't know. I feel like our ability of soft power to just shame them is probably pretty powerful."

Wales said Wikipedia might also consider using technical measures such as Cloudflare's AI Crawl Control, which let clients limit when and how AI bots scraping the internet can access their content. He acknowledged this could create a dilemma, given Wikipedia's ideological commitment to open access to knowledge, but stressed that the financial burden must be addressed.

The Wikimedia Foundation has operated Wikipedia for over two decades as a nonprofit entity, relying on a global community of volunteer editors and public donations to provide free information.

Despite its success, the platform has continuously grappled with maintaining a neutral point of view, particularly on contentious political and social issues. Wales noted that while the vast majority of editors are not activists, it is challenging to maintain calm neutrality during major global conflicts, but that the community "tends to do a pretty good job, even with those circumstances."

Unlock more exclusive Cybernews content on YouTube.