A pilot public dialogue exploring views towards the development and use of AI trained on sensitive data

This report details the findings from a pilot public dialogue led by DARE UK aimed at exploring how to effectively engage people in conversations about AI in sensitive data research, particularly those without existing knowledge and awareness of the subject matter. Three in-person workshops were held with 22 participants in autumn 2025 in Balsall Heath, an area in Birmingham, England. A mixture of presentations, small group discussions and interactive activities enabled participants to develop their understanding of the subject matter and offer informed views. 

Participants expressed optimism about the public benefits of AI trained on sensitive data, but voiced a desire for rigorous data privacy processes, representative training data and clearer lines of consent for data use. Trust was central to discussions, with concern raised about commercial involvement. There was a clear demand that processes to protect sensitive data – including The Five Safes and regulatory requirements – remain fit for purpose in light of the unique challenges posed by AI.

A key challenge throughout the dialogue was maintaining a clear focus on AI trained on sensitive data, with conversations frequently drifting towards internet-based AI such as ChatGPT. This was likely driven by the significant media discourse around mainstream internet-based AI models at the time the workshops took place, and future work will require concerted efforts to maintain a focus on AI in the context of sensitive data research.

DARE UK will be taking the learnings of this work forward to inform a larger dialogue on the same topic with people from across all four nations of the UK. The aim will be to set out more specific, actionable guidance for the development and use of AI trained on sensitive data in line with public expectations.