Security leaders, architects, and teams adopting AI
AI Security
How Henk approaches AI security, secure AI adoption, model risk, governance, and practical guardrails.
AI security is not a bolt-on control set. It is where data, identity, model behaviour, workflow, governance, software delivery, and human trust all meet.
My interest in AI is both strategic and hands-on. I care about how organisations adopt AI safely, but I also build and test AI-enabled systems myself. That practical exposure matters: secure AI adoption is easier to reason about when you understand how the pieces are assembled, how prompts and data flows shape outcomes, and where a system can leak assumptions.
What I Focus On
- Secure AI adoption and operating models.
- Model and data risk.
- Governance that does not collapse into paperwork.
- Guardrails that remain usable under delivery pressure.
- Secure integration between AI tooling, cloud services, identity, logging, and business workflows.
- Evidence that leaders, engineers, and security teams can all understand.
What I Have Done In This Space
I have created and deployed AI and machine-learning learning projects, used programming for security integration and automation, and worked through the security questions that sit around AI-enabled systems. The old site treated some of those projects as demonstrations; this rebuild moves them into the Workbench so they support the professional story rather than dominate it.
The emphasis now is stronger: AI security is a primary professional pillar. The point is not to repeat hype, but to help organisations make AI adoption safer, clearer, more accountable, and more operable.
How This Helps
Good AI security work should answer practical questions:
- What data is being exposed, transformed, retained, or inferred?
- What trust is being placed in the model, the prompt, the integration, and the user?
- What controls are meaningful before, during, and after use?
- What evidence proves the system is behaving acceptably?
- What should stop, degrade, alert, or require human review?
That is where I prefer to work: not theatre, not panic, and not naive acceleration. Clear security thinking for systems that are moving quickly.