ICO responds to the government's AI copyright consultation

The Information Commissioner's Office (ICO) has published its response to the government's Copyright and Artificial Intelligence consultation. 

In its response, the ICO highlights that because some "copyrighted material may also meet the definition of personal data," the government should be aware of and address any potential overlaps between the two regulatory frameworks. In this context, the ICO states that more regulatory clarity may be required to ensure:

• AI developers can collect and process personal data to train AI models, and for;
• Creatives and publishers who need to understand the levers at their disposal to retain control of their published and creative work.

The ICO argues that part of the solution would be for AI developers to provide greater transparency concerning the material used to train models and whether it includes personal data, how the material was acquired, and what content will be generated by these models. The ICO indicates that emphasis on accountability is urgently needed. 

In relation to the proposed opt-out model for text and data mining (TDM) used in the development of AI models, the ICO states that this will not constitute a determination of the lawful basis for any personal data processing under UK data protection law. The ICO goes on to clarify that where any material involved in TDM includes personal data, the "lawfulness of processing would need to be evaluatedon a case-by-case basis."

A blog article by data protection specialist Jon Baines picks up on this point, asking whether the ICO means "each instance" of personal data processing would require a developer to identify a lawful basis or that a developer should identify "a broad, general lawful basis for each dataset." Baines imagines the ICO means the latter. He also questioned whether AI developers would be able to determine the personal data in any given dataset.