Why Jack Thorne Wants Secret AI Scriptwriting to Be Fraud

Jack Thorne has called for laws to ban the secret use of artificial intelligence to generate scripts.

Thorne is the co-writer of the Netflix drama Adolescence and the writer of the stage play Harry Potter and the Cursed Child. He made the remarks in a speech in London as president of the Writers' Guild of Great Britain The Guardian. He warned that some people in scriptwriting are using AI "to cheat". He said writers need to be "squeaky clean" when AI models "steal" creative work.

Thorne said passing off AI-generated work as your own, without declaring it, should be unlawful and treated as fraud. Fraud here means deception for gain. Companies should be entitled to prosecute.

He said broadcasters such as the BBC and Channel 4 should threaten sanctions against writers found to have used AI to generate their work. Sanctions here means penalties, such as loss of commissions or future work.

Thorne also said AI must not be allowed to impersonate actors, directors and writers. He criticised Keir Starmer's government over mass-scraping of material by the AI industry to train models. He said ministers had been "a bit dazzled by tech".

The broader context here is bargaining power. Thorne is linking two disputes that are often handled separately. One concerns inputs, the scraping of existing work to train models. The other concerns outputs, the submission of machine-generated scripts as human authorship. By insisting writers stay clean on outputs, he strengthens the guild's position on inputs. It is similar to handing in someone else's homework with your name on it.

Looking at what this means for policy, the fraud framing does different work from copyright. Copyright arguments focus on permission and payment for training data. Fraud focuses on misrepresentation in the labour market. That gives producers a direct interest in policing disclosure, rather than leaving enforcement to individual creators. The idea is contractual as well as criminal. Disclosure would happen at the point of delivery. Failure to disclose would not be an editorial breach. It would be actionable deception. Enforcement would sit inside commissioning, because broadcasters control access to production and transmission. A credible threat to withdraw that access changes incentives for writers and their representatives. Sanctions by commissioners could operate faster than statute, while legislation would set the baseline.

In my view, the difficulties will be definitional and evidentiary. Script development already involves collaboration, editing tools and iterative drafts. A ban on secret generation requires agreement on what counts as generation, what counts as assistance, and what form of declaration satisfies the law. Prosecution requires proof. Broadcasters would need consistent standards and a process writers trust, or disclosure will be contested case by case. Impersonation raises a parallel set of questions about consent and likeness that sit outside script fees.

To put this in perspective, Thorne's intervention does not resolve those questions. It clarifies the terms the guild wants to fight on: transparency from writers, restraint from developers, and enforcement by the institutions that commission work.