What we know

The Wikimedia Foundation, the organization behind Wikipedia, is advocating for legal protections that would allow large language models (LLMs) to train on copyrighted materials. It warns that unfavorable court decisions in ongoing AI copyright cases could restrict the use of copyrighted content for training, potentially threatening the open web. The Foundation is positioning itself between technology companies and publishers, urging courts to consider the importance of permitting LLMs to access copyrighted works for training purposes. The Wikimedia Foundation warns that restrictive AI copyright rulings could endanger the open web. It supports legal safeguards enabling LLMs to train on copyrighted content.

Why it matters

Several major lawsuits, including one filed by The New York Times against OpenAI, are testing whether AI developers can legally use copyrighted materials to train their models. Although the Wikimedia Foundation is not a direct party in these cases, it has submitted legal briefs highlighting the broader consequences for open knowledge platforms. The Foundation argues that overly strict rulings could undermine collaborative projects like Wikipedia and hinder AI innovation that depends on open access to information.

What is still unknown

The reporting does not provide details on the accuracy of LLMs, the range of languages supported, or any planned release dates related to these legal developments. No independent assessments of the potential impact on AI training or copyright law have been published.