The House of Lords Communications and Digital Committee has urged the Government to rule out any new commercial text-and-data-mining exception with an opt-out model, arguing instead for a licensing-first framework for AI training.
In a report published on 6 March, peers said the UK should compete by building a market for trusted, licensed, and transparent data, rather than by loosening copyright protections in the hope of attracting more model development.
That intervention lands at a critical point in the policy process. The Government said in its December progress statement that it will publish its full report and economic impact assessment on copyright and AI by 18 March 2026. That same statement showed the scale of opposition to the opt-out route floated earlier in the consultation: more than 11,500 responses were received overall, and among Citizen Space respondents 88% backed licensing in all cases, while 3% supported an opt-out exception with transparency measures.
The committee’s central argument is that the UK does not need to rewrite copyright law to support AI. Its report says the present weakness is not an outdated legal framework, but widespread unlicensed use of protected works and limited visibility over how models have been trained. On that basis, peers called for statutory transparency over training data, stronger protections against unauthorised digital replicas and “in the style of” outputs, open standards for provenance and labelling, and greater support for sovereign AI models that can be more readily scrutinised.
Barbara Keeley, chair of the committee, framed the choice in commercial terms as much as legal ones. She said weakening copyright to attract major US technology companies would be “a race to the bottom”, and warned ministers not to trade existing creative-sector value for “AI jam tomorrow”.
That framing matters because the two sectors sit at very different stages of economic maturity. The committee said the UK’s creative industries contributed £124 billion in GVA in 2023 and employed 2.4 million people, with GVA projected to reach £141 billion by 2030. By contrast, the wider AI sector contributed £11.8 billion in GVA in 2024 and employed around 86,000 people. For business leaders, the debate is therefore less about whether AI matters than about which policy setting best supports growth without eroding an already substantial domestic industry.
Peers also pointed to signs that a commercial licensing market is already forming. The report cites a tracker from CREATe showing 120 known agreements globally as of December 2025, spanning publishers, database providers, image libraries, music groups, and media companies. But it argues that a credible UK market cannot be built only around deals between the largest catalogues and the biggest overseas platforms. Any future framework, the committee said, must also work for individual rightsholders, smaller publishers, and small and medium-sized UK AI developers.
The committee acknowledged the counter-argument from parts of the technology sector, which holds that a commercial TDM exception could clarify when licences are needed and encourage more domestic training activity. It said it was not persuaded. Ministers themselves have already moved away from the earlier opt-out preference.
In January, Liz Kendall described the policy debate as a “genuine reset moment”, while Lisa Nandy said it had been “a mistake to start with a preferred model”. With the Government’s report due this month, the remaining question is whether that reset now becomes a durable licensing strategy.
