Definitely hear you on this one Karen! Large Language models don't really reason the way we think (they use complex mathematics rather than reasoning) and companies like OpenAI are shall we say... less than open about their training data. One step towards more transparency we have witnessed is organisations such as Pinnacle PHO conducting Privacy Impact Assessments on tools such as Nabla and making them publicly available here: https://www.pinnaclepractices.co.nz/assets/Resource-files/Nabla-PIA.pdf. I think this would be an excellent topic for a follow on video!