AI and Copyright: House of Lords Recommendations
On March 6, the House of Lords Communications and Digital Committee published a significant report addressing the intersection of AI, Copyright, and the Creative Industries. This report aims to inform the government’s ongoing work concerning copyright and AI, particularly in light of the upcoming consultation report due on March 18.
Committee Recommendations
The report outlines several key recommendations intended to shape future discussions between the creative industries and the AI sector regarding potential reforms to UK copyright law:
1. Rule Out a Commercial TDM Exception
The Committee argues against introducing a Text and Data Mining (TDM) exception with an opt-out model, stating:
- The current UK copyright law is not uncertain, and the act of copying copyright works should not be characterized as “learning,” a common argument from AI developers.
- The government is urged to strengthen licensing, transparency, and enforcement within the existing framework rather than creating new exceptions.
- A final decision regarding copyright and AI should be reached and published within 12 months, establishing strong copyright protection and fair licensing as the default.
- In the interim, a public statement is recommended to clarify that commercial AI developers in the UK must obtain appropriate licenses.
2. Close Gaps in Protection for Style Imitation and Digital Identity
The Committee highlights a significant gap in the UK’s legal framework concerning generative AI outputs that imitate an artist’s style:
- There is currently no specific personality right or protection for digital likenesses.
- The Committee recommends introducing protections against unauthorized digital replicas and uses “in the style of” creators, empowering them with control over their identity’s commercial exploitation.
3. Make Transparency a Statutory Obligation
Granular transparency about the works used in training AI models is crucial:
- The Committee asserts that developers should be mandated to disclose the individual works used in training, despite their objections based on various concerns.
- A clear, mandatory transparency framework should be established, including an appropriate regulator to oversee compliance.
- Care should be taken to avoid imposing excessive burdens on small and medium-sized enterprises (SMEs).
4. Champion the Development of Technical Standards
The report stresses the importance of technical standards for rights-reservation mechanisms:
- These mechanisms are fundamental to any licensing regime, providing granular control for rightsholders while allowing developers to create extensive datasets.
- The government should support the creation and adoption of open, interoperable standards for rights reservation, with a willingness to legislate for effective implementation.
5. Create Conditions for a Fair and Inclusive UK Licensing Market
Evidence presented to the Committee indicates that a licensing market is evolving:
- The UK has the potential to become a global leader in responsible AI development, thanks to its rich creative content and robust collective licensing schemes.
- Collective Management Organisations (CMOs) play a vital role in this ecosystem, and the government should facilitate a sustainable environment for all stakeholders.
Conclusion
One of the report’s most noteworthy conclusions is that current UK copyright law is not ambiguous. The reproduction and processing of digital copies of copyright works for AI training must adhere to established copyright principles. The Committee emphasizes the urgency for the government to publish a statement affirming the necessity for licenses when utilizing copyright works. The prolonged uncertainty surrounding this issue has undermined trust and stalled necessary licensing and investment.
As the government prepares to release its report on copyright under the Data (Use and Access) Act, it remains to be seen how closely it will align with the Committee’s recommendations. The previous favored option of a TDM exception, allowing for an opt-out by rightsholders, may face reconsideration. The upcoming statement could signal a pivotal moment in the relationship between copyright law and AI, setting a clear timeline for decision-making in the year ahead.
