Anthropic has signed its first licensing agreement with a news publisher

Anthropic has officially announced the signing of its first licensing agreement with a news publisher. This decision marks a shift in the company's strategy regarding the use of copyrighted material to train its neural network models.

The deal was concluded just before the U.S. midterm elections. The financial details of the agreement and the name of the partner were not disclosed in the official announcement, but the very fact of moving to a paid content licensing model is a significant precedent for the developer of Claude.

Previously, the company followed a policy of using publicly available data without direct compensation to rights holders. The new agreement indicates Anthropic's desire to minimize legal risks related to copyright and provide a legal basis for training future versions of its models on high-quality news data.

For the AI industry, this move signifies a strengthening trend toward formalizing relationships between generative model developers and media holdings. Publishers now have leverage over AI companies, and developers gain access to verified sources of information, which is critical for reducing hallucinations in model responses.