The Data (Use and Access) Act 2025 received Royal Assent on 19 June 2025 and has brought renewed focus to the relationship between copyright law and artificial intelligence. While the Act itself does not directly amend copyright legislation, it requires the UK government to assess how existing copyright frameworks apply to the training of AI models.
Within the Act, sections 135 to 137, under the heading “Copyright works and artificial intelligence systems”, require the UK government to assess how copyright law applies to the training of AI models. This reflects growing concern that existing legislation has not kept pace with the rapid development of AI technologies, particularly where models are trained on large datasets containing copyright-protected works.
This comes at a time when AI systems rely heavily on large datasets, many of which include copyright-protected material. The challenge is clear: how can the UK support innovation in AI while ensuring that creators retain control over their work?
As a result, the UK has found itself at the centre of a growing debate — how to balance innovation and economic growth with protection of creative rights.
The Consultation
To address this, section 135 of the DUAA requires the Secretary of State to publish an economic impact assessment considering the effect of different policy approaches on copyright owners and AI developers and users. The economic impact assessment must be published within 9 months of when the bill was passed.
On 17 December 2024, the UK government launched a formal consultation on copyright and AI. The consultation closed on 25 February 2025, to gather views on how copyright law should apply to AI training.
The government outlined four policy options:
0. Maintain the status quo – leaving existing copyright rules unchanged.
1. Licensing model – requiring AI developers to obtain permission from rights holders before using their works.
2. Opt-out model – allowing AI systems to use copyrighted works by default unless rights holders actively opt out.
3. Expanded text and data mining (TDM) exception – permitting broader use of copyrighted works for AI training without prior consent.
The consultation gathered 11,512 responses, revealing that 88% of online respondents supports Option 1. Whereas only 7% supported Option 0, 0.5% supported Option 2, and 3% supports Option 3.
Responses revealed a clear divide. AI developers generally supported wider access to data to remain competitive globally, while rights holders argued for stronger safeguards and compensation mechanisms. The opt-out model, in particular, proved highly controversial, with many respondents viewing it as placing an unfair burden on creators to protect their own work.
The Response from the Creative Industry
By early 2026, the consultation had sparked a strong and highly visible reaction from the creative sector. High-profile figures, including Dua Lipa and Elton John, publicly criticised proposals that would allow AI systems to train on copyrighted material without explicit permission.
Their concern was not only legal but economic. Many argued that such proposals risked enabling the uncompensated use of creative works, potentially undermining the value of human creativity and original content.
Industry organisations echoed these concerns, warning that weakening copyright protections could damage the UK’s position as a global leader in music, publishing, and the arts. This level of public and industry pushback became a defining feature of the debate and played a key role in shaping the government’s eventual position.
The Report
In March 2026, in line with the statutory deadline under section 136 DUAA, the government published its report and economic impact assessment on copyright and AI.
Notably, the government did not adopt any of the proposed policy options. Instead, it effectively chose a “no policy change” position for the time being, acknowledging that there is no clear consensus and that further evidence is required.
This represents a shift from earlier indications that an opt-out model might be introduced. Rather than proceeding with legislative change, the government has opted to pause and continue engagement with stakeholders, leaving the current legal framework in place.
In practical terms, this means that existing copyright law continues to apply to AI training, with no new exceptions or permissions introduced as of March 2026.
Impact on the Creative Industry
As of March 2026, the government’s decision has been broadly welcomed by the creative industries, who see it as a temporary safeguard against reforms that could weaken copyright protection.
However, uncertainty remains. Without clear legislative direction, creators still face challenges in understanding how their work may be used in AI training and how to enforce their rights in practice. The debate has also highlighted broader concerns around ∫ which are likely to shape future policy discussions.
Practical Implications
For AI Developers
AI developers remain subject to existing copyright law, with no new exception permitting allowing unrestricted use of protected works.
In practice, this means:
- Continued legal uncertainty when sourcing training data
- Increased reliance on licensed, open-source, or public domain datasets
- Growing pressure to ensure transparency and compliance
While this may slow some aspects of development, it also encourages more responsible and legally robust AI practices.
For Rights-Holders
For rights-holders, the current position provides short-term protection and control.
In practice:
- They retain the ability to license or restrict use of their works
- Existing copyright enforcement mechanisms remain available
- However, uncertainty over future reform continues
Enforcing rights against AI use may still be complex, particularly where datasets are large or where the use is not easily identifiable.
For Consumers
For consumers, the impact is less direct but still significant.
In practice:
- AI tools may become more transparent about how they are trained
- Some AI developments may progress more cautiously
- There may be increased confidence in ethically developed AI systems
Consumers ultimately benefit from both innovation and the continued protection of creative content.
Technology Tuesday Takeaway
The UK government has chosen to pause rather than act. This leaves existing copyright law in place while the debate continues. For now, creators retain control, and AI developers face uncertainty. A s AI continues to evolve, this balancing act between innovation and creative protection is only set to intensify.
Written by Phoebe De Oliveira Simões
24 March 2026