The AI Drop

Megan Kayo, a Freshfields partner focused on AI, cybersecurity, and privacy, joins host Anna Gressel to unpack what makes agentic incidents fundamentally different from traditional cybersecurity incidents. Since agents act on their own, spawn sub-agents, and reason unpredictably, they strain frameworks like NIST built for human threat actors. Megan explains why agents require immutable identities and adherence to the principle of "least agency," and why shadow AI and third-party sprawl make the risk surface so hard to govern. Drawing on her experience guiding clients through incidents, Megan also offers practical steps security and legal teams can take to get ahead of agentic risk.

Download the transcript here

Whether you're automating your morning to-do list or an entire enterprise compliance program, one principle separates success from failure: figure out the process before you hand it to the tool. Suchismita Pahi, Lead Counsel for Product at Databricks, joins host Anna Gressel to trace that thread from her own daily AI workflows to the AI compliance program she built from the ground up . Instead of a legal checklist nobody wants to fill out, Suchi took an operations-first approach that integrates legal, IT, and security. She and Anna dig into what changes as passive LLMs give way to an agentic enterprise — identity management, the principal-agent problem, acceptable error rates - and they share practical insights on building cross-functional AI governance. The episode concludes with a look at the AI use cases that excite Suchi most, including the future of AI-powered scientific research and innovation.

Download the transcript here