Last week, the UK government released its report on copyright and AI, fulfilling a requirement under the Data Use and Access Act 2025 (DUA Act). While the report largely defers policy decisions, it clarifies that the government no longer supports a broad text and data mining (TDM) exception that would have permitted the use of copyright works for training generative AI (GenAI) without an opt-out for rights holders. The report also reiterated the government’s intention to remove existing protections for computer generated works (CGWs) unless evidence emerges proving their ongoing value.
The backdrop for this report includes a December 2024 consultation on AI and copyright initiated by the UK government, which sought to balance the interests of AI developers and intellectual property (IP) rights holders. Among the four policy options presented, the government initially favored a broad TDM exception, allowing for unspecified uses while enabling rights holders to opt out. However, this stance faced significant opposition from the creative industries, resulting in intense parliamentary debate during the DUA Act’s legislative process.
The outcome was a compromise that mandates the government to publish an economic impact assessment and a comprehensive report on the use of copyright works in AI development by March 2026. Last week’s publication encapsulates some key takeaways from this dialogue.
One major finding is the government’s departure from advocating for a broad TDM exception with an opt-out. This decision aligns with earlier acknowledgments that the government had misjudged its position based on the overwhelming feedback received during the consultation—88% of respondents favored strengthening copyright through mandatory licenses, while only 3% supported the government’s initial preference. Although no alternative preference has been proposed, the report mentions possible avenues for further investigation, such as a more focused exception tailored for specific use cases like science and public interest or a broad exception paired with a statutory license akin to models being considered in India.
Currently, the UK maintains a narrow TDM exception limited to non-commercial research. Additionally, the report highlights a widespread consensus on the need for greater transparency regarding the sources used for training data in AI models. At present, there is no obligation for AI developers in the UK to disclose the works utilized for training their GenAI systems. The consultation responses indicated strong support for enhanced transparency, though opinions vary regarding the level of detail and method of disclosure. The government plans to monitor practices in other jurisdictions, such as the European Union and California, in developing best practices for transparency.
The report also addresses UK copyright law concerning models trained outside the country. While the government considered whether UK copyright should apply to these overseas-trained models, it ultimately decided against immediate amendments. Citing potential negative consequences, such as limiting access to models in the UK and adverse effects on local AI developers, the government will defer further action and continue to observe developments in other regions, notably the EU and US, where clarity is anticipated in the near future.
Regarding the licensing market for AI training, the report concludes that government intervention is unwarranted at this time. Instead, the government will keep an eye on the evolving licensing landscape and global trends that may affect it. In relation to CGWs, the UK stands out as one of the few nations providing copyright protection for these works, yet questions about their relevance in the context of GenAI have arisen. The government remains inclined to eliminate these protections unless compelling evidence emerges showing their benefits. The report reflects a prevailing view among respondents supporting the removal of CGW provisions, although the government has chosen to continue monitoring this aspect.
The report also acknowledges rising concerns around unauthorized digital replicas, such as deepfakes, which mimic individuals’ images or voices. With no specific personality or image rights currently established in the UK, the government has proposed exploring options to enhance protections in this area, receiving clear support from consultation respondents for more robust safeguards.
While the government’s findings offer valuable insights into the consultation’s responses and implications of various policy options, many fundamental questions remain unresolved. This ongoing uncertainty surrounding the intersection of copyright and AI is likely to continue, with any legislative reforms pushed into the future. Rights holders may find some solace in the government’s rejection of a broad TDM exception, though the government has yet to dismiss the concept entirely. As the licensing market evolves, the implications of this shift will warrant observation from both the creative and technology sectors, as they navigate the choppy waters of copyright in an AI-driven landscape.
See also
AI Technology Enhances Road Safety in U.S. Cities
China Enforces New Rules Mandating Labeling of AI-Generated Content Starting Next Year
AI-Generated Video of Indian Army Official Criticizing Modi’s Policies Debunked as Fake
JobSphere Launches AI Career Assistant, Reducing Costs by 89% with Multilingual Support
Australia Mandates AI Training for 185,000 Public Servants to Enhance Service Delivery
