A US federal judge has given final approval to a landmark £1.1 billion settlement between AI lab Anthropic and authors and publishers over copyright infringement. The settlement, which was approved on Monday, resolves one case but does not address the broader issue of training AI models on copyrighted works.
The case stems from Anthropic's use of millions of copyrighted books to train its AI models. The company built its training library from two sources: books it purchased and scanned, and books it downloaded from pirate sites. While the settlement delivers £3,000 per work across an estimated 500,000 works, many authors and creators still do not view it as a win.
The legal question of whether training an AI model on copyrighted text counts as fair use was resolved in favour of Anthropic. However, the ruling did not excuse how the company obtained the books in the first place. Anthropic had downloaded millions of books from pirate sites, which was deemed illegal. The company agreed to the settlement to avoid a trial and potential damages.
The final approval of the settlement closes out this case but does not settle the legal question industry-wide. Other judges are still free to reach their own conclusions on the issue, which is playing out in other copyright lawsuits against companies such as Google, Meta, Midjourney, and OpenAI.
The issue of training AI models on copyrighted works is still ongoing, with a string of copyright lawsuits against major tech companies. Just last week, a group of publishers and authors filed a class action lawsuit against Google over accusations that the company used their copyrighted works to train its AI platform, Gemini.