On the same day, before publishing the reports, Liz Kendall, Secretary of State for Science, Innovation and Technology (DSIT) made a written statement to Parliament setting out the government’s commitments going forward (Copyright and AI – Hansard – UK Parliament).
Key outcomes
Liz Kendall was clear in her written statement that the government will not rush into anything because “it will take time to get this right”. However, she identified four areas where the government will focus the next phase of work: digital replicas, labelling AI-generated content, creator control and transparency and launching a working group of independent creatives. In addition to these, she highlighted a government initiative to establish a creative content exchange.
It is clear from the Report that the 11,000 plus Consultation responses and subsequent technical working groups have not cut a clear path through the grass. There is no immediate reform for AI and copyright on the cards, with the text and data mining (“TDM“) exception and opt-out mechanism, previously favoured by the government, no longer the preferred option. Instead, the status quo continues with the government opting to monitor ongoing litigation (e.g. Getty Images v Stability AI), international developments and market activity before committing the UK to a specific course of action on AI and copyright. The Report emphasises the government’s aim of protecting the UK’s position as a “creative powerhouse” while “unlocking the extraordinary potential of AI to grow the economy”. Key focuses for the future highlighted by the Report include further engagement with industry and experts on licensing, input transparency, industry-led standards and AI output labelling.
There are other interesting developments arising from the Report. The government is inclined to remove the computer-generated works (“CGW“) provision in the Copyright, Designs and Patents Act 1988 (“CDPA“). As Liz Kendall highlighted in her written statement, the government also plans to look more closely at digital replicas and what sort of regulation is required for them – potentially considering the introduction of a personality right in the UK.
Recap on the December 2024 Consultation
The UK government launched its highly anticipated consultation on copyright and AI, (which we wrote about here: UK plans to be an “AI maker” not an “AI taker” with its new copyright consultation and AI Opportunities Action Plan and here: 2025 will be ‘a blockbuster year’ for AI) seeking views on the following four proposals:
- Option 0: Do nothing with copyright and related laws remaining as they are (“Status Quo Option“).
- Option 1: Strengthen copyright requiring licensing in all cases (“Strengthen Licensing Option“).
- Option 2: Introduce a new, broad data mining exception (“Broad TDM Option“).
- Option 3: Introduce a new, broad data mining exception with an opt-out mechanism (“TDM and Opt-Out Option“).
At the time, the UK government preferred the introduction of a new TDM exception for commercial purposes with rightsholders permitted to use an opt-out/rights reservation to control content use with the ambition of being a global AI leader (while respecting its creative industries).
Other areas considered in the Consultation were licensing, transparency obligations for AI developers training AI models, AI content labelling, CGW and digital replicas.
The March 2026 Report and Impact Assessment
Report on Copyright and AI
The Report addresses the responses to the different Consultation proposals as follows:
- TDM and Opt-Out Option: The headline for many will be the government’s reversal on the TDM and Opt-Out Option which is no longer the preferred option. Consultation responses highlighted the practical and economic difficulties of opt‑out regimes and international experience (notably in the EU) has reinforced those concerns.
- Strengthen Licensing Option: This option was the most supported option from the Consultation (but note that most respondents were from the creative industry). Many respondents viewed it as broadly consistent with the current UK copyright framework, which already requires licences in most commercial AI training scenarios. Support focused less on changing substantive rights and more on making licensing workable in practice, especially through stronger transparency from AI developers. However, the government requires more evidence on how it works in practice and will also monitor developments in international markets.
- Status Quo Option: There was some support for the Status Quo Option with creative industry respondents pointing out that existing law already enables licensing and warning that reform could disrupt a nascent AI‑training licensing market, stressing that developers should continue to seek permission. AI developers and users disagreed, arguing the current regime hampers AI development and leaves the UK more restrictive than jurisdictions like the US, increasing cost and risk. The risk for SMEs and individuals in particular was noted by respondents across sectors.
- Broad TDM Option (no opt-out): This was the least favoured option with 3% of respondents supporting the Broad TDM Option, many of those coming from the AI, technology and research sectors who thought this option was a compromise that addressed the competing interests and helped achieve alignment with the EU.
Other key themes arising from the Report are examined below:
Transparency
On training data, the Report acknowledges the widespread frustration on lack of transparency.
Over 90% of respondents, from both the creative and AI sectors, agreed that AI developers should disclose the sources of their training data. The real disagreement is not over whether transparency is needed, but how far it should go and whether it should be regulated.
Creative industry respondents strongly backed legislative or regulatory transparency, seeing it as essential to understanding copyright use, enabling licensing, supporting enforcement, and building trust. Many favoured an EU‑style, market‑access approach, applying transparency obligations regardless of where an AI system is developed, while others supported industry codes backed by enforcement tools such as audits and financial sanctions.
AI developers, including SMEs, pushed back on work‑by‑work disclosure, warning it would be costly and impractical. They favoured high‑level, industry‑led transparency instead of mandatory regimes. Views were mixed on following the EU model: developers saw alignment benefits, while creatives questioned its effectiveness.
Transparency is described as essential to enforcement, licensing and trust. However, the government will not legislate at this stage. Instead, it intends to track developments under the EU AI Act, US initiatives and emerging international standards on data disclosure, including clearer crawler identification and source information.
Liz Kendall indicated in her written statement that DSIT will publish a review of the mechanisms available for creators to control their works online, to include standards, technical solutions and best practice on input transparency. The review will inform where there are gaps and whether there is an appropriate role for government in addressing them.
Enforcement
On enforcement, the Report concludes that the UK’s legal framework is fundamentally robust but hampered by limited visibility over training data and the inherently cross‑border nature of AI model development.
The government does not intend to create a new regulator. It plans to work with industry, enforcement bodies and the courts to ensure effective and accessible routes to redress, including for AI systems developed outside the UK, while weighing impacts on both rightsholders and AI innovation. It will track enforcement models overseas (notably the EU) and build an evidence base on what works for AI development and rightsholder remedies.
Licensing
The Report explored different approaches to licensing, noting that no statutory licensing regimes were identified (although acknowledging that the Indian government has recently consulted on one). The strong overall message from creative and AI respondents is that the government should not pursue legislative licensing (on a direct or collective licensing basis). This should continue to be managed between commercial parties. The creative sector was keen for the government to focus on creating market conditions that encouraged licensing with a key focus on transparency obligations for AI developers to enable better licensing and enforcement.
The government is not pursuing statutory or compulsory licensing for now, arguing that the market is developing quickly. It will continue monitoring collective licensing initiatives and the broader international landscape.
Technical tools and standards
The Report highlights the rising importance of machine‑readable controls—robots.txt, metadata, permission signals and other technical mechanisms that help manage the use of copyright works. While the government sees clear value in these tools, it does not propose mandating a single standard, preferring to encourage interoperability and market‑led adoption.
Liz Kendall indicated in her written statement that DSIT will publish a review of the mechanisms available for creators to control their works online, to include standards, technical solutions and best practice on input transparency. The review will inform where there are gaps and whether there is an appropriate role for government in addressing them.
AI output labelling
Those who responded on AI output labelling broadly supported it to mitigate risks such as deepfakes and misinformation. An important distinction was drawn between wholly AI‑generated content and AI‑assisted works involving human input with the former requiring labelling but not the latter.
There was support for some regulatory underpinning, reflecting concern that voluntary approaches have produced fragmentation. Suggestions included automatic labelling via watermarks or metadata, with some favouring obligations limited to particular platforms such as social media or news services.
Again, however, the government is avoiding immediate intervention, preferring industry‑led guidance and alignment with international models once they settle.
Liz Kendall indicated in her written statement that DSIT will establish a taskforce to put forward proposals on best practice for labelling AI-generated content, with an interim report to be published in the autumn.
CGWs
The Report also confirms strong support for removing protection for CGWs under s.9(3) of the CDPA. The government proposes doing so, subject to further evidence (it proposes to continue to monitor the use and impact of this protection), while maintaining the existing originality test for AI‑assisted works.
Digital replicas
Finally, the government acknowledges a significant gap in UK law concerning digital replicas and personality rights—including synthetic voice, image and likeness. Existing legal tools provide incomplete protection, and the government will explore potential new rights or models, though no definitive approach is adopted yet.
However, Liz Kendall indicated in her written statement that the government will launch a consultation in the summer to seek views on how to address harms caused by digital replicas, while protecting legitimate innovation.
Impact Assessment
The accompanying Impact Assessment evaluated the economic implications of the four policy options set out in the Consultation.
The Impact Assessment adopts no preferred option and confirms that the government will not make immediate legislative changes. A consistent theme across all scenarios is the transparency deficit, which continues to undermine both enforcement and licensing. The Impact Assessment also highlights the particular vulnerability of SMEs and individual creators, who may struggle to engage with or influence large‑scale AI development and associated licensing markets.
Crucially, the Impact Assessment concludes that international developments are likely to shape the UK’s practical trajectory more quickly than domestic reform. The implementation of the EU AI Act, coupled with ongoing US litigation, is expected to drive commercial expectations and market behaviours in the short term.
In short, the Impact Assessment reinforces the government’s overarching message – this is a moment for strategic pause, not immediate overhaul.
Comparison with the report from the House of Lords’ Communications and Digital Committee (“Committee”)
On 6 March 2026, the Committee published a detailed and fairly critical report on the effects of generative artificial intelligence (GenAI) on the creative industry which we wrote about here – Lords Committee report warns of Generative AI risks to creatives.
The Committee warns that the UK creative industries face a “clear and present danger” from GenAI trained on copyright-protected works without authorisation or remuneration. It concludes that the problem lies not with UK copyright law, which it describes as an international “gold standard”, but with widespread unlicensed use and lack of transparency by AI developers.
Key messages from the Committee’s report included:
- Firm rejection of any commercial TDM exception, including opt‑out models.
- A licensing‑first approach to AI development, supported by statutory transparency obligations.
- A call for mandatory, granular disclosure of AI training data sources, overseen by a regulator and applying to all models placed on the UK market, regardless of where they are trained.
- Identification of serious gaps in protection for digital replicas, voice, image and “in the style of” outputs, with recommendations for new enforceable rights.
- A demand for Government clarity within 12 months, warning that prolonged uncertainty is already stalling licensing and investment.
The Committee is explicit that weakening copyright in pursuit of speculative AI gains is neither necessary nor justified, given the creative industries’ economic significance and the harms already being experienced by creators.
The government’s Report marks a reset from its earlier preference for a TDM exception, aligning with the Lords on several core issues: UK copyright law is not broken; a broad TDM exception is no longer the preferred option; transparency and licensing are central to a sustainable AI ecosystem; and market‑led licensing should be supported.
However, the government adopts a much more cautious approach – rather than legislating, it has opted for a pause, focusing on evidence‑gathering, monitoring international developments (notably the EU AI Act and US litigation), and commissioning further research. It has declined, for now, to impose statutory transparency obligations, create a new enforcement regulator, or commit to a concrete timetable for reform.
Comment
The UK government has chosen to play the long game on AI and copyright, focusing on evidence‑gathering as it cites significant gaps and uncertainty in how the AI and copyright market is developing. It also intends to commission research into the impact of copyright reform on the creative industries and AI sector, draw on the findings of the DSIT AI sector study (published in September 2025) and continue with its stakeholder engagement to better understand how developers source data, use TDM in practice, and to assess the extent to which copyright affects business activity.
Creatives will welcome the news that a broad TDM and opt-out model is no longer the preferred approach. However, it was hoped that these publications would offer more clarity on some of the most contested issues in the copyright and AI debate and identify the UK’s path forward. Many of the key proposals set out in the Consultation have been put on hold while the government gathers yet more evidence, watches developments in other jurisdictions and assesses impacts.
The rationale for the government’s “watching brief” is understandable but the lack of a clear policy direction is likely to be met with concern by many who had hoped that the government would provide a more concrete plan of action more than a year after publication of the Consultation (and 5 years since the initial consultation on IP and AI in 2021).
Maintaining the status quo means that for now, rightsholders, creators and AI developers in the UK will have to continue navigating uncertain waters. There is currently no easy route for creatives to identify when and what works are being used, to seek adequate remuneration for use of their works or to enforce their rights. Across the aisle, AI developers face challenges accessing works on a licensed basis and the uncertain litigation outlook represents a considerable business risk. As a result, they may prefer to conduct AI training, research and development activities in more permissive jurisdictions such as the EU and the US, which would hamper the UK’s ambition for the UK to be “one of the best places in the world to build and adopt innovative AI”.
With thanks to Charlotte Fleetwood-Smith, Senior Associate, and Jessica Gurzynski, Solicitor Apprentice in Fieldfisher’s IP team, for their contribution to this article.
