Baroness Harding of Winscombe (Con)

My Lords, I will also try to be brief. I completely support everything that the noble Baroness, Lady Kidron, has said. I would like to draw out two arguments that have been made to me today as to why her amendments should not be supported and explain why they are wrong. The first argument is that we should wait for an overarching AI Bill. We will be waiting for a very long time. Those of us who have worked in trying to regulate social media for the last 15 years know that we must not let the perfect be the enemy of the good. I wholeheartedly reject that argument. The second argument that has been made to me today—and I find this astonishing—is that the risk assessment is overly burdensome. We are regularly told that generative AI is one of the world’s most transformational technologies. That means it is capable of enormous good and enormous harm. The risk assessment in Amendment 433 is simply asking that the makers of these chatbots identify and understand the risks of harm—that does not seem overly burdensome to me. Further, it asks that the risk assessment “is kept up-to-date … takes … account … of the Online Safety Act … assesses the risks to equality of treatment of individuals … assesses the risks to … privacy … assesses the risks … from the choice of underlying models, data sets …and … is in an easily understandable written format”. I really struggle to understand how that could be overly burdensome. In fact, I would argue the absolute opposite: it is the basic foundation of decent regulation, and we should be wholeheartedly supporting the amendments from the noble Baroness, Lady Kidron.