Gaurav Keerthi, CEO of StrongKeep, a Singapore-based cybersecurity startup that provides SMEs with AI-driven security and compliance platforms, returns to Kopi Time to discuss the topic of the moment. In this wide ranging conversation, we begin by talking about why secure AI is a fundamentally different proposition compared to traditional software. Gaurav walks us through the essence of an AI-driven large language model, the unintended consequences that can arise even without hallucination or bad actor involvement, and the layers of risks involved. We comb through recent headlines to underscore the quick changing environment around cyber safety and security. We then take stock of the resources needed to deal with this constantly evolving risk, and how that leaves particularly small businesses vulnerable. Gaurav calls for cybersecurity in the AI era to be considered a public good that informs all business and regulatory decision-making. He is also keen to see a fundamental reshaping of cyber infrastructure, with built-in, publicly funded and monitored guardrails and reporting. The silver lining? The well-intentioned efforts toward innovation and safety can keep pace with the pernicious aspects. There is hope.