Okta just paid $200M for a company that monitors AI agents, not humans

Okta just paid $200M for Permiso Security.

Not for human identity management.
For securing AI agents that have standing access to your infrastructure.

This is the first major identity vendor to acquire dedicated AI agent threat detection.
Permiso monitors 70+ identity providers using 2,500+ behavioral signals to detect when AI agents do something they shouldn't.

The numbers behind this deal are terrifying.
Non-human identities outnumber human users 45 to 1 in enterprise environments.
In cloud-native setups, that ratio hits 144 to 1.
Only 15% of organizations feel confident preventing attacks on these identities.
16% don't even track when new AI-related credentials are created.

Okta's own data shows 58% of executives reported an AI-related security incident or near miss in the past year.
Permiso's SandyClaw platform detonates AI agent skills in a sandbox before they reach production.
That's not a nice-to-have.
That's a requirement.

Your CISO just got a new line item.
The NHI access management market is projected to grow from $11.14B to $27.33B by 2033.
Okta is positioning itself to own that category before CrowdStrike or Palo Alto Networks does.

Audit your non-human identity governance today.
If you cannot tell me how many AI agents have standing privileges in your environment, how many API keys are unrotated, and which third-party integrations have never been reviewed, your identity perimeter is already compromised.

This is the new attack surface.
And Okta just told you the market believes it's worth $200M to solve.