AI agents are moving beyond simple tools, and organizations need to rethink how they manage them.
In Part 2 of Trust Issues, Bruno Lecoq and Jeremy Patterson, Co-host and Account Executive at BEMO, continue their conversation with Jamal Ahmed, Founder and CEO of Kazient Privacy Experts, about the realities of AI governance and what responsible adoption looks like in practice.
They explore why AI agents need clear ownership, identity management, access controls, and regular reviews, much like human employees. They also discuss the danger of confidently wrong AI outputs and why people cannot use “AI told me” as an excuse when something goes wrong.
The conversation then turns to what it really takes to build a career in AI governance. Jamal explains why experience does not automatically equal competence, why multiple-choice certifications cannot prove that someone can do the job, and how practical learning and real-world artifacts help professionals demonstrate their abilities.
They also examine the four characteristics that make AI different from traditional software: it learns, it is opaque, it is probabilistic, and it changes over time. Finally, Jamal discusses the growing importance of the EU AI Act, NIST AI Risk Management Framework, and ISO 42001, and why professionals need to understand how these frameworks work together.
What You’ll Learn:
- Why AI agents need clear ownership, reporting lines, and regular reviews
- How identity and access management can help organizations control AI agents
- What multiple-choice certifications demonstrate, and what they cannot prove
- How practical learning, mentorship, and real-world artifacts help build AI governance competence
- Why saying “I don’t know” can be a responsible answer when assessing AI
- The four characteristics that make AI different from traditional software
- Why AI governance requires ongoing assessment rather than a one-time review
- How the EU AI Act, NIST AI Risk Management Framework, and ISO 42001 fit into the wider regulatory landscape
Episode chapters: [00:00] Governing AI Agents Like Employees at Scale
[01:21] Identity and Access Management for AI Agents
[02:35] Confidently Wrong: The Danger of Overconfident AI
[04:16] The Driving Theory Test Analogy for AI Governance
[07:03] Why Experience Doesn’t Equal Competence
[08:30] Inside the Certified AI Governance Expert Program
[12:19] ISO 42001 and Vendor Compliance Requirements
[13:43] Four Traits Every AI System Shares
[14:48] The EU AI Act, NIST AI RMF, and ISO 42001
[15:48] Embracing AI Without Losing Your Guard
[17:32] Data Privacy Hygiene When Using AI
[18:53] Key Takeaways and Final Thoughts
Quotes:
“An agent is always confident in his answer, and even when it’s wrong, it’s confidently wrong.”
“Experience for competence is a misconception. You could be doing something really badly for a long time; that doesn’t make you competent.”
“There is a lot of stuff that is opaque in AI governance, and saying ‘I don’t know’ or ‘I need this information to answer’ is actually a perfectly acceptable answer.”
“AI learns, it is opaque, it is probabilistic, and it’s going to change over time.”
“Embrace AI. Whether we like it or not, you’re going to get left behind if you don’t.”
“Don’t put something into an AI tool or system that you wouldn’t be giving out in Times Square to a stranger in New York.”
Connect with the team: 👉 Jamal Ahmed LinkedIn: https://www.linkedin.com/in/kmjahmed/
👉Kazient Privacy Experts Website: https://kazient.co.uk/#dataeasy
👉 Jeremy Patterson LinkedIn: https://www.linkedin.com/in/jeremypatterson2026/
👉 Bruno Lecoq on LinkedIn: https://www.linkedin.com/in/brunolecoq/
👉 BEMO Website: https://www.bemopro.com/ Trust Issues is handcrafted by our friends over at: fame.so