Baroness Kidron (CB)

My Lords, I thank the noble Baroness, Lady Morgan of Cotes, and the noble Lords, Lord Russell and Lord Clement-Jones, for adding their names. I also thank the noble Baroness, Lady Barran, for trying to add her name. Such was the enthusiasm that there was no space. As already discussed, the Government have brought in new Clauses 92 and 93 to allow companies and responsible third parties to risk-assess the creation of CSA by gen AI models. That is an important detail. If the company is red teaming, or the regulator needs to test, it must not be guilty of an offence for doing so. But this new measure is permission, not obligation—and permission is not enough. Amendment 209 seeks to do three things: to make risk assessment mandatory; to require mitigation within 14 days; and to hold companies not covered by the Online Safety Act to the same standard via the National Crime Agency. A report from UNICEF last month referenced an Interpol study across 11 countries which found that at least 1.2 million children have disclosed having their images manipulated into sexually explicit images in the past year. In some countries that is equivalent to one child in every classroom being subjected to this new form of child sexual abuse. The report recommended the introduction of guardrails for AI developers at the design stage. In a meeting earlier in your Lordships’ House, we were told repeatedly and reminded graphically that AI CSAM creates appetite in offenders and that what happens online does not stay online. We have consulted, and Ofcom has consulted—Parliament has debated this for years—and now we are consulting again. I argue that there are three reasons for accepting the amendment right now.