Music publishers allege that Anthropic violated copyright policies by using hundreds of songs to train its Claude chatbot.

A partial settlement signed in the U.S. District Court of Northern California requires Anthropic to use “guardrails” to protect copyrighted lyrics allegedly used to train its premier AI product, Claude.

The AI company still argues that training Claude on copyrighted material constitutes fair use.

Artificial intelligence company Anthropic has reached an agreement with Universal Music and other music publishers over its use of "guardrails" to keep its chatbot Claude from generating copyrighted song lyrics,

Anthropic and the record labels that sued it over copyright concerns agree the Amazon-backed AI firm's products can't recite protected music or make derivative work.

While the case is proceeding, Anthropic's Claude will no longer provide lyrics to songs owned by the music publishers or new lyrics based on copyrighted material.

The music industry's first victory over Anthropic: the AI provider must install guard rails. However, the question of training remains open.

The deal requires Anthropic to ensure Claude doesn't use or generate lyrics from copyrighted songs. This unique case, the first where a music publisher has sued an AI company over