On March 10, about 10,000 writers collectively participated in the publication of a special work called “Don’t Steal This Book” to protest against artificial intelligence companies using their works as training data without permission and payment. The book has almost no text and mainly just lists the names of all participants, including well-known authors such as Nobel Prize winner Kazuo Ishiguro, best-selling historical novelist Philippa Gregory, and crime novelist Richard Osman.
This "blank book" was distributed for free to attendees at the London Book Fair, just in time for the British government's announcement of an economic impact assessment of changes to copyright law in a week's time. According to the arrangement, the government needs to submit to Parliament before March 18 an economic impact assessment of the proposed copyright reform, as well as an updated progress report on previous consultations on relevant legal adjustments. Amid strong dissatisfaction among creative industry practitioners over the use of copyrighted works by AI companies, this issue has sparked ongoing debate in the British cultural and technological circles.
Ed Newton-Rex, the organizer of the action, composer and activist, pointed out that the current generative AI industry is "built on stolen works" and related content is "used without permission and payment." He emphasized that this was not a "victimless act" because generative AI products directly competed in the market with the works of creators on which their training relied, "depriving creators of their livelihoods" and called on the government to protect British creative workers and "refuse to legalize the 'theft' of AI companies."
Participating in adding their names to the book are Mick Herron, author of the "Slow Horses" series, best-selling author Marianne Kees, historian David Olusoga, and Mallory Blackman, author of the "Noughts and Crosses" series. Blackman said that asking AI companies to pay for the use of authors' works is "not excessive at all." The declaration on the back cover of the book reads: "The British government must not legalize theft of books to facilitate AI companies."
At the same time, the publishing industry also launched new AI licensing initiatives at the London Book Fair. The non-profit industry organization “Publishers’ Licensing Services” announced that it will establish a collective licensing mechanism and invite the publishing industry to join, with a view to providing AI companies with a legal way to obtain published works. The move is seen as an attempt to find an institutional solution between protecting copyright and supporting technological innovation.
Generative AI models require massive amounts of data during development, including large amounts of copyrighted content scraped from the open web to train tools such as chatbots and image generators. This practice has caused widespread concern among creative practitioners and content companies around the world, and has spawned a number of transatlantic lawsuits. Last year, Anthropic, a well-known AI company and developer of the chatbot Claude, agreed to pay $1.5 billion to settle a class-action lawsuit initiated by multiple book authors, who accused the company of using pirated books to train its core model without authorization.
One of the most controversial proposals in the British government's current consultation on copyright reform is to allow AI companies to use copyrighted works for training without the copyright owner explicitly "opting out." In other words, AI use is authorized by default. Once the copyright owner does not want the work to be used as training data, he must actively object. This "tacit consent, then withdrawal" design aroused strong indignation among British artists and creative groups. Many artists, including musician Elton John, have spoken out against relaxing copyright protections, and John even denounced the government's approach to AI copyright plans as a "complete loser."
In addition to the above main plan, the government also proposed three alternative options in the consultation text: First, keep the current system unchanged; second, require AI companies to obtain permission and pay fees before using copyrighted content for training; third, allow AI companies to fully use copyrighted works without any "exit mechanism" - this option is regarded by the creative industry as the most radical "comprehensive exemption" plan.
Furthermore, the government has not ruled out granting copyright exemptions for “commercial research” purposes. Creative practitioners are worried that once the exemption clause for "commercial research" is written into law, AI companies may use this as a pretext to continue to capture and use the works of artists, writers and other groups on a large scale without explicit authorization, thereby further weakening the protection of creators' rights.
In response to questions from various parties, a spokesperson for the British government responded that the government hopes to establish a copyright system that "values and protects human creativity, is trustworthy, and promotes innovation." He promised to continue to communicate closely with the creative industry and fulfill its obligation to update Parliament on relevant progress before March 18. The political and legal battle surrounding the copyright ownership, compensation mechanism and licensing model of AI training data is becoming a key front in the cultural and technological fields in the UK and even around the world. This "almost blank" "Don't Steal This Book" is one of the collective signals the creators are trying to send.