The Test AI Can’t Fail: Building the Lines of Defense for Child Safety

Join us for an evening of drinks, light bites, and discussion on one of the most urgent questions facing AI safety teams today: How can organizations improve child safety and prevent AI being exploited to generate illegal content? As generative AI systems become more capable, GenAI companies need ways to confidently ensure their models cannot produce harmful or illegal content, including Child Sexual Abuse Material (CSAM), Non-Consensual Intimate Imagery (NCII), extreme pornography (EP), and other severe abuse risks. But testing for these harms creates its own challenge: the work often requires teams to prompt models to generate the very content they are trying to prevent. Moderated by Hive, this panel will bring together leaders from government, advocacy, and industry, including the UK Home Office, Thorn, and others, to discuss the safeguards needed to support AI child safety. The conversation will explore how teams are approaching internal red teaming, third-party testing, proactive detection tools, and emerging legal and policy frameworks, including initiatives like TAKE IT DOWN…