United Kingdom — AI Copyright Law

A Unique Approach: Protecting "Computer-Generated Works"

While jurisdictions like the United States and the European Union struggle to fit AI outputs into traditional frameworks demanding human authorship, the United Kingdom has possessed specific legislation addressing this issue for decades.

Section 9(3) of the CDPA 1988

The Copyright, Designs and Patents Act 1988 (CDPA) contains a unique provision regarding works generated by computers where no human author exists.

Section 9(3) of the CDPA states: "In the case of a literary, dramatic, musical or artistic work which is computer-generated, the author shall be taken to be the person by whom the arrangements necessary for the creation of the work are undertaken."

Section 178 further defines "computer-generated" as a work "generated by computer in circumstances such that there is no human author of the work."

Who Makes the "Necessary Arrangements"?

This phrasing creates a fascinating dynamic for generative AI. If a user prompts an AI like Midjourney or ChatGPT, who made the "necessary arrangements"?

• The User? Because they provided the specific prompt and parameters that triggered the creation.
• The AI Developer? Because they built, trained, and provided the infrastructure without which the work could not exist.
• Both? Potentially as joint authors under certain interpretations.

While early case law (e.g., Nova Productions v Mazooma Games, regarding video games) suggested the programmer was the author, the advent of modern generative AI shifts more creative agency to the prompter. Current legal consensus is leaning toward the user who crafts complex prompts being the person undertaking the necessary arrangements, thereby securing a 50-year copyright term (shorter than the standard life-plus-70 years).

This makes the UK currently one of the most favorable jurisdictions for businesses seeking to commercialize AI-generated assets, as it provides a statutory basis for ownership that is explicitly denied in the US.

AI Training Data and Text/Data Mining (TDM)

The UK's approach to the input side of the AI equation—training models on copyrighted data—has been turbulent and marks a significant divergence from the EU post-Brexit.

The Current TDM Exception (Non-Commercial Only)

Currently, Section 29A of the CDPA provides an exception for Text and Data Mining, but it is strictly limited to non-commercial research. This means commercial AI developers cannot rely on it to scrape the UK web for training data without licenses.

The Abandoned Broad Exception

In 2022, the UK government announced plans to introduce a broad, commercial TDM exception that would allow AI training on any accessible data without the possibility for creators to opt-out. This aimed to make the UK a global hub for AI development.

However, following intense backlash from the creative industries (publishing, music, and visual arts), the government completely abandoned this proposal in early 2023. This leaves commercial AI developers in the UK in a precarious position compared to those in Japan or the EU.

The UK AI Code of Practice

Following the abandonment of the broad TDM exception, the UK Intellectual Property Office (IPO) attempted to broker a voluntary "Code of Practice" between AI developers and the creative industries.

The goal was to establish guidelines for licensing data and providing transparency. However, negotiations collapsed as the sides could not agree on fundamental principles regarding licensing and opt-outs. The government is currently exploring legislative options to mandate transparency and facilitate licensing markets, moving closer to a regulatory approach rather than relying solely on voluntary industry agreements.

Keep track of ongoing IPO consultations and regulatory shifts in our UK Policy Tracker.

Comparison Summary

Feature	United Kingdom	United States	European Union
Copyright for purely AI-generated works	Yes (via "necessary arrangements" clause, 50-year term)	No (Strictly requires human authorship)	Generally No (Requires "author's own intellectual creation")
Commercial AI Training Exception	No (Non-commercial research only)	Relying on "Fair Use" litigation	Yes (But with an explicit right for creators to opt-out)
Regulatory Approach	Sector-specific guidelines, light-touch regulation	Litigation-driven, executive orders	Comprehensive, heavy ex-ante regulation (AI Act)

Practical Takeaways

• For Content Creators: If you use AI heavily, the UK offers a safer harbor for claiming ownership of the resulting outputs compared to the US, thanks to Section 9(3).
• For AI Developers in the UK: You cannot currently scrape UK-based copyrighted content for commercial training without severe infringement risks. You must seek explicit licenses from rights holders or rely on public domain datasets.
• For Global Businesses: The divergence between UK and EU law post-Brexit means compliance in Europe requires two distinct strategies—one for the EU's structured opt-out system, and one for the UK's stricter, license-first environment for training data.