What changed with the UK’s AI decision on 19 March?

The government scrapped its preferred opt-out approach for training on copyrighted works. There is now no single preferred option, so developers must rely on licences, permissions, or clearly lawful sources. This increases uncertainty and strengthens rights holders’ leverage over fees, terms, and transparency across the UK market.

How could this affect AI developers’ costs in the UK?

Licensing and compliance costs may rise. Teams may pay for curated datasets, negotiate collective or direct licences, and invest in provenance audits, labelling, and takedown processes. Some models could require retraining or dataset swaps, adding time and GBP expense to UK deployments and updates.

What should investors monitor next in the UK?

Watch government consultations on labelling, deepfakes, and licensing frameworks. Look for signals on collective licensing, disclosure duties, and safe-harbour ideas. Track early court cases and settlements that set price and process norms. These factors will drive margins, time-to-market, and scale for UK-focused AI providers.

Does this affect models already trained on UK content?

Yes, potential exposure exists if datasets included copyrighted UK material without clear rights. Outcomes will depend on licences, provenance records, and any future rules or rulings. Firms may need to adjust datasets, secure permissions, or retrain parts of models to reduce legal and commercial risk.

Law and Government

UK U-Turn on AI Copyright, March 19: Licensing Risk for Big Tech

ByDanny Kontos March 20, 2026

AI copyright UK moved centre-stage on 19 March as the government scrapped its preferred opt-out plan for AI training on copyrighted works. The shift removes a clear route for text-and-data mining and raises near-term licensing and compliance risk. Rights holders gain leverage, while AI developers face uncertainty on labelling, deepfakes, and fees. For UK investors, this is a pricing and margin story: data costs could rise, models may need retraining, and timelines could stretch as firms adjust to UK AI regulation and tighter oversight.