🎧 Spotify
Straight from Black Hat, Diana Jovin, CMO of Teleport, takes the hot seat — tackling rapid-fire questions on Agentic AI, Non-Human Identities, and the future of identity security in autonomous environments. Listen to how Diana responded to the questions below:
Are CISOs Ready for the Fleet of Autonomous Agents That Are About to Hit Them?
Diana explains why agents behave differently from both humans and machines — and why their speed and autonomy, colliding with existing environments, creates real risk of them doing things you never intended. Her verdict: teams need to get ready now.
Is IAM for AI More Important Than AI for IAM?
Diana pushes back on the framing itself. From Teleport’s infrastructure-first perspective, the real question isn’t IAM-for-AI versus AI-for-IAM — it’s how you let your first-class actors (agents, humans, machines, workloads) interact securely. She reframes it as an infrastructure question, not an IAM one.
Are Organisations Deploying Agents Into Production Without Appropriate Guardrails?
Diana doesn’t hesitate: “of course.” She explains why the consequences are already visible, and why deployment speed is once again running ahead of governance.
Do AI Agents Behave More Like Employees or Software?
Diana’s answer: neither. She explains why two decades of zero-trust principles built for humans and machines don’t fully transfer — using the example of an agent told to update a database that’s down and choosing to go find another one — and why agents need controls mapped to how they actually behave.
Are the Current Identity Standards Fit for Autonomous Systems?
Diana argues we need a new foundation: no static credentials, no standing privileges, no static digital representations of identity for agents to discover and exploit. She lays out a two-part model — a strong zero-trust base plus governance mapped to agent behaviour.
Does PAM Change Dramatically With Agentic AI?
Diana frames PAM as a concept built for human identity management — and argues we’re now in a new era its assumptions were never designed for.
Will Static Long-Lived Credentials Still Exist in 5 Years?
“I certainly hope not.” Diana makes the case for getting long-lived credentials out of infrastructure altogether.
Do AI Agents Need Proper Background Checks Like Employees?
Diana reframes the question around delegated identity: every agent — and everything it spawns — should carry a strong identity with attestation back to its origin. If the origin was vetted, the agent inherits it; what you should never do is create an agent outside those controls.
Does Agentic AI Security Equal NHI Security?
Diana argues words matter. She explains why “NHI” has come to mean static representations of machine identities that need to disappear, why agents behave differently from machine workloads, and why the field needs a clearer shared vocabulary rather than a single equivalence.
Will the AI Digital Workforce Lead to Mass Unemployment?
A firm “no.” Diana expects AI to redefine how people work rather than replace them wholesale — opening new opportunities for transformation we haven’t uncovered yet, with this just the start of a longer wave of disruption.
NHI Foundation Level Training
If you work with AI agents, cloud services, APIs, or CI/CD pipelines, securing Non-Human Identities is essential.
Take the most comprehensive and only accredited NHI training program in the industry and learn how to manage and secure NHIs, including AI agents.
Enrol here – nhi-training.org
