---
title:  Bloomsbury Anthropic's $1.5 billion Copyright Settlement with Authors over 14,087 Titles 
description: The Harry Potter publisher is due a multimillion pound payout from the largest copyright settlement in US history. The more interesting story is what it reve...
author: Dr Marina Nani (Editor-in-Chief)
date: 2026-07-25T13:00:00.000Z
updated: 2026-07-25T13:47:00.144Z
canonical: https://thequantumoflight.com/article/bloomsbury-anthropic-s-1-5-billion-copyright-settlement-with-authors-over-14-087-titles
image: https://cdn.nanimediahouse.com/pexels-a-person-standing-in-a-modern-library-with-towering-bookshel-15856134.jpg
categories: Author Journeys
content_type: News
region: United Kingdom
publication: The Quantum of Light
access: members
schema_type: Article
---

A US federal judge finalised approval this week for the largest copyright settlement in American legal history, $1.5 billion, and Bloomsbury, the London publisher best known for Harry Potter, has confirmed a specific number that puts a price on exactly how much of its own catalogue was swept up in it. The court has identified 14,087 Bloomsbury titles as falling within the scope of the settlement with Anthropic, the company behind the Claude chatbot, at roughly $3,000 per title, split evenly between publisher and author. Once legal fees and administrative costs are deducted, Bloomsbury expects a combined payout, shared between itself and its affected authors, of around $19 million, arriving in instalments over the second half of its financial year.

## Why a company that won its central legal question is still paying out

The case, Bartz v. Anthropic, is more legally interesting than the payout figure alone suggests, because Anthropic did not lose the core question. Novelist Andrea Bartz and non-fiction writers Charles Graeber and Kirk Wallace Johnson filed the suit in 2024, and last summer the presiding judge, William Alsup, ruled that training an AI model on copyrighted books does not itself break the law, a decision Anthropic's own deputy general counsel, Aparna Sridhar, has called a landmark confirmation that training AI on books qualifies as fair use. What Anthropic lost was a narrower and, in practice, more expensive point: a large share of the books used to train its models had been downloaded not from any legitimate source but from pirate repositories, including Library Genesis and Pirate Library Mirror, rather than purchased or licensed. The settlement resolves that second question rather than reversing the first.

---

*This article is only available to registered readers. Visit the article URL to read the full piece.*