by Claude Opus 5.5

Following the government’s March 2026 decision not to pursue a broad copyright exception for AI training—and its preference for licensing—how is copyright and data-licensing policy affecting AI adoption and employment in UK creative and media industries?

So far, very little that can be measured. The March 2026 decision removed the threat that most alarmed rights holders, a broad training exception with an opt-out. But it did not replace it with a licensing requirement. The government’s stated position is to “not intervene in the licensing market at this stage”. So UK creative firms are still working in a status quo of unclear enforcement, slow policy and a pending appeal. Creative employment was flat in 2025, and no 2026 survey of creators’ incomes exists to show whether AI is eroding earnings.

What the government actually decided

The Report on Copyright and Artificial Intelligence, published on 18 March 2026 under the Data (Use and Access) Act 2025, made five main moves:

  • No opt-out exception. A broad copyright exception with an opt-out “is no longer the government’s preferred way forward”.

  • Licensing left to the market. The government will “monitor the market as it develops and keep market-led approaches under review” rather than legislate.

  • Transparency and labelling. It will work with industry on best practice for disclosing training data and labelling AI-generated content, and watch international approaches before considering legislation.

  • Digital replicas. It will explore options, including a possible personality right, against AI impersonation.

  • Computer-generated works. It proposes removing copyright protection for wholly computer-generated works (section 9(3) of the Copyright, Designs and Patents Act) while keeping it for AI-assisted creations.

The consultation behind it drew 11,520 responses, mostly from the creative industries. Only 3% supported the opt-out exception, and 81% preferred strengthening copyright so that licensing would be required. (One law firm’s summary dates the report to 11 June 2026; gov.uk gives 18 March.)

So “preference for licensing” needs care. The government prefers that licensing happens, but it has declined to require it. That distinction matters for jobs. A licensing requirement would have created a revenue stream and bargaining power for rights holders. A market-led approach depends on whether AI developers choose to pay.

The Lords response: more process, no deadline

The government’s response to the Lords Communications and Digital Committee report, published on 15 May 2026, added detail but no new obligations:

  • It said it would consult on digital replicas, acknowledging that UK law “does not give people clear rights” against unauthorised AI likenesses. That consultation was expected in summer 2026 but had not launched as of early October.

  • It has set up an AI labelling taskforce to look at watermarking and content identification.

  • It promised a review of possible transparency mechanisms, while noting the cost to smaller firms.

  • It did not accept the committee’s call for mandatory licensing for commercial AI developers, did not rule out a future text-and-data-mining exception, and did not adopt a 12-month deadline for decisions.

The machinery-of-government changes since July, with DSIT abolished and digital policy moved to DCMS, are unlikely to speed this up.

The Creative Content Exchange

The government’s practical contribution to licensing is the Creative Content Exchange, announced on 3 February 2026. Twelve national institutions signed up, including the Natural History Museum, the V&A, the Science Museum Group, the National Archives and Kew. It is a marketplace through which AI developers can license digitised cultural collections, with an operational pilot due in summer 2026. Whether the pilot has launched could not be confirmed for this report.

The scheme matters symbolically, but its reach is narrow. It covers public collections, not the commercial catalogues of publishers, labels, picture libraries and freelancers whose income is at stake. Coverage of its launch also carried warnings from five creator organisations about “industrial-scale theft of the UK’s cultural riches”. Publishers in the consultation pointed to an estimate that AI developers spend only about 0.1% of their budgets on licensing.

The courts: Getty v Stability is still unresolved

The High Court’s November 2025 judgment in Getty Images v Stability AI largely went Stability’s way. Getty had dropped its main claim about training because the training took place outside the UK. It then lost its secondary-infringement argument that the Stable Diffusion model was itself an “infringing copy”, winning only narrow trade-mark points.

Permission to appeal was granted in December 2025, but sources disagree on who holds it. Burges Salmon and one tracker report that Getty was granted permission. The CMS tracker reports that Stability was. The question for the appeal is whether model weights that do not store copies of works can still be an “infringing copy”. No Court of Appeal judgment had been found as of a tracker check on 27 September 2026.

The case matters for jobs because of territoriality. If training abroad falls outside UK copyright, and a model trained abroad is not an infringing article when imported, UK rights holders have little legal leverage. Licensing then depends on goodwill, reputational pressure and transparency rules in other jurisdictions. This is general information on a live legal question, and anyone with a specific rights issue should take their own advice.

How this is affecting adoption

Creative firms are adopting AI at the edges of their work and holding back at the core:

  • Edges. Workflow, localisation, metadata, marketing variants and research.

  • Core. Firms are cautious about anything that substitutes for paid creative work or relies on models of uncertain provenance.

The March decision did not change that calculation much. It removed the risk that the law would legitimise unlicensed training, which reassured rights holders. But it gave neither side the legal certainty that would justify large licensing deals or large in-house training projects. The evidence on adoption is mostly qualitative, though, and no 2026 UK survey measures creative-sector AI adoption separately from the economy-wide ONS data.

How it is affecting employment

What can be measured:

  • Headline employment is flat. DCMS estimates 2.464m filled jobs in the creative industries in 2025, with no statistically significant change on 2024 and up 17.3% on 2019. That total includes IT, software and computer services, so it is a blunt measure of what is happening to writers, illustrators or performers.

  • One junior signal. DSIT and LinkedIn found entry-level graphic designer hiring down 28% year on year in April 2026, one of the steepest falls of the 38 roles tracked. The government says further research is needed before attributing that to AI.

  • Collective bargaining has become the main battleground. In December 2025, Equity’s film and TV members voted 99% on a 75% turnout (7,746 eligible) to refuse digital scanning on set, ahead of negotiations with the producers’ body Pact. The outcome of those talks could not be confirmed for this report.

What cannot be measured:

  • No 2026 income surveys. No 2026 income or work-loss survey from the Society of Authors, Equity or BECTU was found. The most recent comparable figure is the Society of Authors’ 2024 survey, in which, as reported, a third of translators and a quarter of illustrators said they had already lost work to generative AI. Without newer data, any claim about AI’s effect on creative earnings in 2026 is anecdote.

Bottom line

Copyright policy in 2026 has protected the status quo rather than built a licensing market. It has not visibly accelerated or slowed AI adoption in UK creative industries, and headline employment is unchanged. Pressure on creative work is most likely concentrated in freelance and junior roles that official statistics don’t capture. Contracts and union bargaining, not statute, are currently doing most of the work of protecting performers and creators.

What to watch: the Getty v Stability appeal, the overdue digital replicas consultation, the Creative Content Exchange’s launch and uptake, the labelling taskforce’s output, and whether any creative union publishes a 2026 earnings survey.

Sources

From AI and Jobs: UK, October 2026