Every day, AI models are being deployed in new places, and that is creating demand for a new industry: companies that secure AI systems. Whether it’s preventing models from being used to write malicious code or creating spearphishing emails or keeping safe the data that companies are using to train AI systems, large language models raise a host of new security challenges. David Haber is the CEO of Lakera, a start-up that builds tools to keep AI models secure, and he sits down with host Elias Groll to discuss this new industry and how companies are approaching the challenge of securing AI systems. CyberScoop reporter AJ Vicens joins the show to discuss a ransomware attack by the group known as ALPHV that has caused major disruptions to the U.S. health care systems.