What happens when artificial intelligence moves faster than our ability to understand, verify, and trust it?
In this episode of Tech Talks Daily, I sit down with Alexander Feick from eSentire, a cybersecurity veteran who has spent more than a decade working at the intersection of complex systems, risk, and emerging technology. Alex leads eSentire Labs, where his team explores how new technologies can be secured before they quietly become load-bearing parts of modern business infrastructure.
Our conversation centers on a timely and uncomfortable reality. AI is being embedded into workflows, products, and decision-making systems at a pace most organizations are not prepared for.
Alex explains why many AI failures are not caused by malicious models or dramatic breaches, but by broken ownership, invisible dependencies, and a lack of ongoing verification. These are not technical glitches. They are organizational blind spots that quietly compound risk over time.
We also explore the ideas behind Alex's recently published book on trust and AI, which he made freely available due to the speed at which real-world AI failures were already overtaking theory.
From prompt injection and model drift to the dangers of treating non-deterministic systems as if they were predictable software, Alex shares why generative AI requires a fundamentally different security mindset. He draws a clear distinction between chatbot AI and embedded AI, and explains the moment where trust quietly shifts away from humans and into systems that cannot take accountability.
The discussion goes deeper into what trust actually means in an AI-driven organization. Alex argues that trust must be earned, measured, and monitored continuously, not assumed after a successful pilot. Verification becomes the real work, not generation, and leaders who fail to recognize that shift risk scaling errors faster than they can contain them. We also talk about why he turned his book into an AI advisor, what that experiment revealed about the limits of models, and why human responsibility cannot be automated away.
This is a grounded, practical conversation for leaders, technologists, and anyone deploying AI inside real organizations. If AI is becoming part of how decisions get made where you work, how confident are you that someone truly owns the outcome?
Useful Links
Tech Talks Daily is sponsored by Denodo
