Philip Pullman urges UK action on AI scraping: "It's a wicked system"
Sir Philip Pullman wants the government to close the gap between ethics and law on AI training data. He says writers' work is being scraped to build AI models without pay or credit - and that needs to stop.
"They can do what they like with my work if they pay me for it," he said. "But stealing people's work... and then passing it off as something else... That's immoral but unfortunately not illegal."
His stance echoes concerns from other authors, including Kate Mosse and Richard Osman, who warn that unlicensed scraping could drain value from creative work and hollow out future growth.
Where policy stands now
Last December, the UK launched a consultation on copyright and AI across the creative industries, attracting 11,500 responses. Expert working groups on AI and copyright are now in motion, with the Department for Culture, Media and Sport engaged.
For context on the legal direction, see the UK IPO's consultation on copyright and AI and past calls for views:
- Artificial intelligence and intellectual property: copyright and patents (UK Government)
- AI and IP policy overview (WIPO)
What government can do now
- Create a clear, enforceable licensing pathway for text and data mining of books and scripts - default to opt-in, not opt-out.
- Require AI developers to disclose training datasets, sources, and licensing status, with penalties for concealment.
- Establish collective licensing options with transparent revenue distribution to authors and publishers.
- Mandate audit trails for model training so rightsholders can trace usage and claim payment.
- Set a fast-track dispute process for takedowns, injunctive relief, and statutory damages on willful infringement.
- Fund independent testing to detect model outputs that imitate identifiable living authors.
- Publish a timetable from the current working groups to draft, consult on, and implement changes.
What writers and publishers can do
- Audit contracts and permissions to cover data mining, synthetic voice rights, and derivative outputs.
- Work through collecting societies and trade bodies to push for collective bargaining and standard licenses.
- Lobby MPs with concrete asks: dataset disclosure, opt-in licensing, audits, and statutory damages.
- Track AI outputs that mimic your style; document evidence and seek legal advice early.
- Update site policies and metadata signals on scraping. They won't stop bad actors, but they clarify your stance.
Pullman's broader point: protect imagination
Pullman's new book, The Rose Field, closes his second trilogy about Lyra Silvertongue. Across six books that have sold 49 million copies, Lyra confronts a power structure (the Magisterium) that fears imagination and tries to suppress it.
He argues curiosity should be praised in children, not muted by rigid schooling that prizes rote learning and silence in corridors. He also criticises the use of religion by people in power to force compliance.
Release details
The Rose Field: The Book of Dust Volume Three will be published on 23 October. The audiobook is read by actor Michael Sheen. Lyra was portrayed by Dafne Keen in the BBC series that ran from 2019 to 2022.
The line in the sand
Pullman is blunt: "Of course they should change it at once. Don't you steal anybody else's work." The ask is simple - compensate writers, disclose datasets, and stop laundering unlicensed work through AI.
If the policy response matches the clarity of that principle, the creative economy keeps its engine - and AI can grow on fair terms, not free rides.
Your membership also unlocks:
