Anthropic, an AI lab, has received final court approval for its $1.5 billion settlement in a class action lawsuit over copyright infringement, allowing the company to distribute payments to affected authors and publishers. The settlement covers roughly 500,000 copyrighted works, with each work earning $3,000 to be shared among its rights holders. This agreement follows a legal battle initiated after Anthropic was found to have downloaded millions of copyrighted books without authorization.

The controversy stemmed from Anthropic’s method of compiling training data, which included both legitimately purchased books and those acquired from pirate websites like Library Genesis. While a previous judge ruled that training AI on copyrighted texts constitutes fair use—a significant decision favoring the AI industry—the same judge also condemned the illegal downloading as copyright violation. To avoid a trial and potential damages, Anthropic agreed to the settlement, which was ultimately approved by a succeeding judge after the original jurist retired.

Though this settlement resolves the specific case involving Anthropic, it does not establish broad legal precedent for the AI industry's use of copyrighted materials in model training. Since the case will not advance to an appeals court, other courts remain free to interpret copyright law differently in similar disputes. This uncertainty is underscored by ongoing lawsuits against major AI and tech companies such as Google, Meta, Midjourney, and OpenAI.

Adding to the legal turmoil, a new class action was recently filed against Google by several prominent publishers and authors, alleging unauthorized use of their copyrighted works in training Google's AI platform, Gemini. As AI development advances rapidly, the legal landscape around copyright and AI training data remains unsettled, with this landmark Anthropic settlement merely marking one step in a larger, ongoing industry debate.