Claroty: AI Agent Guardrails & Non-Human Identity Controls

Name: Claroty: AI Agent Guardrails & Non-Human Identity Controls
Uploaded: 2026-08-23T09:28:46-04:00
Duration: 1 min 35 s
Description: This video features John Laliberte, CEO of ClearVector, discussing the critical need for security guardrails around AI agents and non-human identities. He highlights how traditional Identity Providers (IdPs) focus on human credentials, leaving non-huma...

Claroty

08/23/2026

0 (0%)

Report Like Favorite

TL;DR

Traditional IdPs like Okta and Entra were designed for human identities, leaving AI agents and non-human identities with uncontrolled direct access to enterprise environments.
Security guardrails — analogous to bowling bumper rails — are needed to let AI agents operate quickly while constraining them to expected, authorized actions only.
A critical blind spot exists when employees hand their own credentials to an LLM, making it impossible to tell whether account activity is human or AI-driven without specialized detection.

Summary

In this short clip from the Nexus Podcast, John Laliberte, CEO of ClearVector, outlines a pressing security challenge: as AI agents proliferate inside enterprise environments, traditional Identity Providers (IdPs) like Okta and Microsoft Entra were built exclusively around human credentials — leaving non-human and AI agent identities with direct, unmonitored access to sensitive systems. Laliberte uses a bowling bumper analogy to explain the concept of guardrails: controls that allow teams to move fast with AI agents while constraining those agents to only expected, authorized behaviors. He notes that IDP vendors are beginning to codify non-human identity provisioning, but a critical gap remains — employees can simply hand their own credentials to an LLM, which then operates autonomously under a human identity. ClearVector addresses this by detecting whether observed account activity reflects genuine human behavior or an LLM acting on a person's behalf, helping security teams distinguish between legitimate AI delegation and potential adversarial misuse of human credentials.

Chapters

0:00 - The Guardrails Concept
0:21 - IDP Vendors Respond
0:51 - The Credential Delegation Problem
1:15 - ClearVector's Detection Approach

Key Quotes

0:12 "How can you allow people to move fast with agents, but also make it so agents can only do what you expect them to do, or at least have kind of a bounding box around them."
0:30 "Non-humans are not in your IDP, right? They have direct access to these environments."
0:55 "You can just give your credentials to an LLM or an agent, and it's gonna operate with your credentials."
1:04 "Is the activity that we're seeing from this identity still human? Or do we think an LLM is operating with that human's account and it's actually AI behind the scenes? ..."

FAQ

Why can't existing IdPs like Okta or Entra handle AI agent identities today?

Traditional IdPs were designed around human credentials and provisioning workflows. Non-human identities — including AI agents — typically have direct access to environments outside of IDP governance, meaning the same policy controls applied to human accounts do not extend to them.

What happens when an employee gives their credentials to an AI agent?

The AI agent operates under the employee's human identity, making its activity indistinguishable from normal human behavior without specialized detection. ClearVector claims to identify this shift — flagging when an account transitions from human click-based activity to LLM-driven behavior.

Categories:

» Cybersecurity » Zero Trust
» Data Protection

Tags:

Show more Show less

TL;DR

Summary

Chapters

Key Quotes

FAQ

Becoming Agent Ready with Cyera: Essential Strategies and Insights

Verge.io: Can You Afford Your Next Storage Refresh?

Summer of Satori: FunFoneFarm's Transformation of Fraud into Seamless Integration

Invisible Data: Understanding What Needs Protection

Embrace AI Adoption While Maintaining Robust Security Measures