How to Build AI Systems People Can Actually Trust | Cillian Kieran, CEO & Founder, Ethyca

In this episode, Cillian Kieran, CEO and founder of Ethyca, joins Liam to talk about why data privacy and AI governance need to be treated as engineering problems, not just legal and compliance challenges.

Cillian explains how a consulting project for Heineken ahead of the GDPR pushed him to rethink privacy from an engineer's perspective and eventually build Ethyca. He breaks down how the company's product suite works, why Fides has become a widely adopted open-source privacy standard, and what changes when AI agents are given read and write access to tools like Stripe, QuickBooks or a CRM.

They also get into what foundation model providers may be missing on governance, why AI needs a harness that directs its capabilities without slowing it down, the responsibility engineers have when building AI systems, and why he thinks much of what we call AI is still statistical math wrapped in a marketing label.

Later, Cillian and Liam discuss where AI startups may consolidate, Yann LeCun's work on world models, the human impact of increasingly agent-driven work, and how growing up around art shaped Cillian's view of software as a creative pursuit.

Key Topics Covered

Why privacy and AI governance are engineering problems

How Fides became a widely adopted open-source privacy standard

The risk of giving AI agents read and write access to business systems

Why AI needs a harness that directs its capabilities without slowing it down

What foundation model providers may be missing on governance

Why engineers, not just users, carry responsibility for AI safety

Why Cillian calls AI "statistical math" with a marketing label