The UK passes the Data (Use and Access) Bill, criticized by creatives, without an amendment to force companies to declare using copyrighted material to train AI
A bill which sparked an extraordinary stand-off between some of the UK's most high-profile artists - and their backers in the House of Lords - has finally been passed.
Context & Ripple Effects
The bill closes a parliamentary fight in which prominent artists and House of Lords backers sought a disclosure requirement for AI training on copyrighted material. It follows a period in which the UK had shelved planned rules for AI training on copyrighted material after creative and technology groups failed to reach terms.
The dispute was already widening beyond individual artists: a [[a:880513|coalition of publishers, writers and film producers opposed a proposed AI copyright exemption]]. Subsequent coverage of the government's withdrawal of an opt-out AI-training proposal shows that copyright and training-data policy remained unsettled after this bill.
First-order effects
- Companies do not face the proposed disclosure obligation under the Data (Use and Access) Bill when using copyrighted material to train AI.
- Creative professionals lose a legislative mechanism they sought to identify and challenge possible use of their works in AI training.
Second-order effects
- Absent a disclosure mandate, rights holders have less information with which to assess licensing, attribution or infringement claims, increasing the importance of voluntary transparency and private agreements.
- The unresolved issue keeps pressure on the UK government to address AI copyright separately, rather than treating this bill as a final settlement of training-data rules.
Third-order effects
- The episode points to a persistent gap between AI developers' demand for scalable training inputs and creators' demand for traceability and control; closing it may require a more explicit data-rights framework.
- If policy continues to proceed in piecemeal bills and withdrawn proposals, the UK may see rules shaped through sector-specific negotiations and disputes rather than one durable AI-copyright settlement.
The trend: AI governance is increasingly turning on the provenance, permission and disclosure rules for the data used to build commercial models.