More companies join Wikimedia Enterprise for AI training

Amazon, Meta, Microsoft, Mistral AI and Perplexity formalized their relationships with the platform through Wikimedia Enterprise

Wikipedia gained support from five more companies that use its content to train their artificial intelligence platforms.

Amazon, Meta, Microsoft, Mistral AI and Perplexity formalized their relationships with the platform through Wikimedia Enterprise, the Wikimedia Foundation’s commercial product for large-scale reusers and distributors of content from the foundation’s projects, Wikimedia Enterprise said in a Thursday blog post.

These companies join the ecosystem’s existing partners, which include Google, Ecosia, Nomic, Pleias, ProRata and Reef Media, according to the post.

Wikipedia, which is marking its 25th anniversary on Thursday, has become “one of the highest-quality datasets for training large language models,” the post said. The platform’s 65 million articles in 300 languages help power generative AI chatbots, search engines and voice assistants.

In the AI era, Wikipedia’s human-created and curated knowledge has never been more valuable, Wikimedia Enterprise said in the post.

Wikimedia Enterprise provides tech companies with high-throughput API access to Wikipedia through their choice of an On-demand API that returns the most recent version for an article request, a Snapshot API that provides Wikipedia as a downloadable file that is updated every hour, and a Realtime API that streams updates as they are made, according to the post.

As knowledge on Wikipedia continues to grow to close knowledge gaps and include more languages, its value as a dataset for a broad spectrum of use cases also increases, the post said.

Perplexity said in a Wednesday blog post that to mark Wikipedia’s anniversary, it is offering the platform a gift of 2,500 Perplexity Enterprise seats to share with its editors.

Perplexity is a Wikimedia Enterprise customer because we’re proud to support the foundation’s mission to keep knowledge free and accessible for everyone, the company said in its post.