‘Pay up or get out’: UK Lords want AI giants to licence creative content before training, not after
The UK’s House of Lords says that if companies like OpenAI and Google want to train their models on British creative content, they need to pay for licences upfront.
In a report published this month by the Lords’ Communications and Digital Committee, lawmakers paint a stark picture of two possible futures for the UK’s AI and creative sectors.
One is a “world-leading” ecosystem where AI developers secure permissions and “pay fair remuneration” to rightsholders before scraping content for training.
The other is a drift toward “tacit acceptance” of unlicensed data hoovering, with most gains flowing to a handful of US tech giants while UK creators suffer.
“Only the first path is compatible with the UK’s long-term interests. The UK’s creative industries are an economic powerhouse that contributed £124 billion to the UK economy in 2023, with gross value added expected to reach £141 billion by 2030,” the report states.
“Their success is underpinned by a ‘gold-standard’ copyright framework, which rewards creativity, supports sustainable business models for creative work, and commands international respect.”
The 85-page document comes amid growing tensions between AI innovators and content creators worldwide. Lawsuits from authors, artists, and publishers against AI firms have piled up in the US and Europe, accusing them of mass copyright infringement through unauthorised data mining.
The Lords’ inquiry, which drew on evidence from stakeholders across tech and media, urges the UK government to reject calls for looser copyright rules that would allow broad “text and data mining” (TDM) exceptions for commercial AI training.
Such a move, the report warns, would “remove incentives to license protected works” and inflict “predictable harms” on rightsholders.
Instead, it pushes for a “licencing-first approach,” backed by technical tools to let creators opt out of AI scraping and mandatory transparency from developers about what data they’ve used.
The economic stakes are huge
The UK’s creative industries – spanning film, music, publishing, and design – are an economic juggernaut, pumping £124 billion into the economy in 2023, with projections hitting £141 billion by 2030.
That’s far outpacing the relatively new AI sector, which the report pegs at a more modest £3.7 billion in gross value added and 50,000 jobs, per 2023 figures.
But generative AI, capable of churning out text, images, and music in seconds, is built on “vast quantities of human-created content, much of it copyrighted,” the Lords note.
Without proper safeguards, creators risk losing control over their work, facing “tangible economic harms” like undercut commissions and an influx of cheap AI-generated alternatives flooding the market.
The report slams the current “limited transparency” from AI firms, which leaves rightsholders in the dark about whether their material has been ingested.
It also highlights gaps in UK law, such as the lack of robust protections against AI imitating a creator’s style, voice, or digital identity – issues that have sparked outrage among performers and artists.
Key recommendations
To tilt the scales toward fair play, the report outlines a multi-pronged strategy:
Boost transparency: AI developers should disclose training data sources, with granular details where possible. The committee floats mandatory rules, potentially aligned with the EU’s AI Act, which requires summaries of copyrighted data used in high-risk systems.
Tech fixes for rights protection: Endorse “rights-reservation tools” like site-level blocks (e.g., robots.txt) and asset-level tags to signal opt-outs. But it warns these alone aren’t enough, advocating for them as a “foundation for a licensing-first approach.”
Supercharge licensing: Foster an “emerging AI licensing market” through collective schemes that benefit individual creators, not just big publishers. The government should play matchmaker, perhaps via a “Creative Content Exchange” to streamline deals.
Sovereign AI push: Encourage homegrown, transparent models to reduce reliance on “opaque models trained overseas.”
The report nods to Australia’s recent stance, where the government clarified that AI firms must comply with existing copyright laws and seek permissions.
What this means for AI giants – and the UK
For the US-based behemoths dominating generative AI, the message is clear: “Pay up or get out” of using UK content without deals.
OpenAI has already inked licensing pacts with publishers like Axel Springer and the Associated Press, but the Lords want this as the norm, not the exception.
Google, Meta, and others could face tougher scrutiny if the UK adopts these ideas, especially with territorial rules applying to models deployed in Britain.
By championing “responsible, licensing-based” AI, the report argues that UK could carve out a niche in “creative technology,” blending its creative prowess with tech innovation.
But inaction risks “long-term dependence” on foreign AI, eroding domestic edges.
