A month after IMPACT, the energy from the Idira launch is still here. The customer conversations have not slowed down for a day. The hype is real, and it is earned.
Two moments stand out.
The first was Nikesh Arora, our CEO, framing identity as the next inflection point in cybersecurity, created by AI. He called out something the industry has been quietly working around for a decade: the hyper-fragmentation of identity into siloed boxes. PAM for the privileged workforce. IAM for the broader workforce. NHI for the new things. ISPM if you set it up wrong. IGA, if you can’t manage it.
These little boxes have instructions on how to stay inside them. Except, as he put it, some of these identities live across all the boxes. (It’s well worth watching the IMPACT keynote).
The second standout was Peretz Regev, introducing Idira and the premise underneath it. Every identity is privileged, and AI introduces autonomous, machine-speed action that makes this impossible to ignore.
The question every CISO and identity leader I talked to that week was the same: What does this look like on Monday morning?
“Democratizing” privilege controls means extending the rigor of privileged access management to every identity that now carries privilege: Workforce users, workloads, machines, and AI agents.
IAM and PAM: What Stays, What Has to Change
I don’t want to live in a world where we declare the last decade of privilege work a failure. It was not. The PAM generation built vaulting, session isolation, just-in-time elevation, recording, and lifecycle controls for the highest-risk accounts and environments. Those controls should stay. What has to change is the reach.
The boxes stopped working the moment a single workflow started running across all of them.
- A workforce user can change a production configuration through a SaaS console.
- A developer can deploy code that changes production.
- A workload can pull from a database that the whole quarter depends on.
- An AI agent can do all three on someone else's behalf.
Most access attempts in any enterprise are legitimate. We don’t need an admin-style vault and approval flow in front of every one of them. The fix here isn’t adding another identity category to the pile. It’s platformization. Democratizing privilege takes the rigor we built for the smallest, riskiest population and applies it to every population that now carries privilege. Keep what works. Extend it to everyone who needs it.
The Three “S”s of Democratized Privilege Management
The way to evaluate any democratized privilege model, including ours, is whether it delivers on three things at once:
- Scale. The controls must reach every identity that now carries privilege via one control plane, not five. The 109:1 machine-to-human ratio is the sizing reality every privilege control surface has to handle.
- Speed. Privilege decisions run at machine speed, not at the agonizing speed of an approval queue. Just-in-time (JIT) access becomes the default cadence, not the exception. Continuous evaluation, not quarterly access review. The control has to move at the speed of the attacker, which is faster than the speed of your inbox.
- Security. The strongest privilege control available is scoped to what the identity is doing, only when the action calls for it and without disrupting work. Zero standing privilege (ZSP) is the default for the whole identity surface.
What Changes for Humans: Managing Workforce Privilege
A regular workforce user today has access to systems that can cause real damage if abused. They can change a production configuration through a SaaS console. They can prompt an AI assistant to pull customer data into a working document. They can push code through a low-code workflow that runs in production. They can approve a vendor onboarding. That’s a different risk from the one the IAM model was built for.
Let's be direct: privilege is no longer something you can pin to a job title or to membership in a group.
The same user can run a low-risk task in the morning and a high-impact one in the afternoon. The control has to understand the difference without sitting on top of every click. What matters is the action, what it touches, what assistant or agent is involved, and whether it warrants a privilege guardrail.
For the team building this, the work shifts from modeling who someone is to modeling what they are doing—and on whose behalf.
Same Principles, New Identities: Machines, Workloads, and AI Agents
Privileged access management taught the industry how to handle high-risk identities. Discover what exists. Apply tight controls when access is granted. Track what happens. Pull access back when the work is done. Those principles are right. They were just applied to a small slice of the identity surface.
This is what it looks like when we extend these principles to every non-human identity.
- Machines and Workloads. The static secret sitting in a config file or a vault entry is the old shape. The new shape is workload identity tied to attestation, with credentials generated for the specific call the workload is making and removed once it completes. The same strong control, but now running at the speed and scope of the actual work, not at the speed of a quarterly key rotation.
- AI Acting on Behalf of a Human. Today, most enterprises use Copilot or Claude inside the productivity stack. Agent features ship inside SaaS that the security team did not procure. The AI acts inside a human's session, using a slice of the human's access, touching more data in a single action than the human ever would on their own.
Democratized privilege means the assistant gets a scoped, observable, revocable piece of that access. It can’t escalate beyond the human's access, and the privilege check happens at the call, not at the login. Most identity programs have not modeled this yet. It is the immediate work.
- Autonomous AI agents. The next layer. Their own identity, their own scoped privilege, their own audit trail. The teams that handle on-behalf-of well today are the ones who will get autonomy on their terms. The ones who ignore it inherit autonomy with no controls in place. 99% of organizations already use autonomous AI agents, and most of those agents use human or machine credentials that were never sized for this kind of work.
The Operating Model That Makes Identity Security Hold at Scale
The admin who built the PAM program is the load-bearing wall of this entire story. If their role does not change, none of the above scales. The controls reach further, the queue gets longer, the team burns out, and the program quietly retreats to admin-only privilege a few quarters later.
The shift is from gatekeeper to architect: from approving each request to designing the system that makes approval decisions at machine speed. The work moves from operational throughput to operational design across three principles that the team now owns end-to-end. Discover, control and govern.
| Discover | Maintains a continuous view of every identity, its entitlements, and the access paths it can reach. | Replaces quarterly inventory scans with a living view across human, machine, and agentic identities. |
| Control | Applies the right privilege control to the right action at the right time. | Makes ZSP and just-in-time access the default across workforce users, developers, workloads, and agents. |
| Govern | Records, reviews, and acts on every privilege decision, assignment, revocation, and exception. | Turns compliance from a quarterly assembly exercise into a continuous state of the platform. |
Without this shift, every new developer, workload, and agent lands back on the same admin queue. The 12-hour fragmentation tax does not go away. It compounds.
This is the hardest change in the program. It’s also the one that starts paying back the day it begins.
Idira: The Unified Identity Security Platform
For most of the last decade, identity has been an integration project. PAM, IAM, NHI, ISPM, IGA. Each tool was handed over with instructions to make them work together. That model does not survive a 109-to-1 identity ratio.
Idira is the identity security platform built to discover, control, and govern privilege across human, machine, and AI identities. It sits alongside Strata and Cortex as the third platform pillar at Palo Alto Networks, and the integration work is the platform's job, not yours.
When you look across your architecture, this native platform consolidation translates directly into preconfigured, out-of-the-box automation where privilege lives today:
- Cortex receives first-party identity signals to sharpen detection and trigger an identity-driven response with Idira.
- Prisma Browser puts privileged access controls where enterprise users already work.
- Prisma AIRS 3.0 extends these controls to AI agents through native integration with Idira.
- Next-Generation Trust Security, which we launched at RSA earlier this year, automates certificate lifecycle at the network layer.
The team running the program does not spend its week stitching together a view of identity events from five disconnected systems. The view is already there.
This Is a Today Decision
Putting this off does not buy time. Workforce users have access to systems that can cause real and immediate damage. Machine identities vastly outnumber humans, and AI assistants are running inside the productivity tools that shipped last quarter, with or without security team signoff. Every one of those identities is privileged, but the controls around them are still the ones built for the admin tier. AI accelerates every part of this. Now, attackers can move from foothold to exfiltration in 72 minutes.
The same safeguards that were once reserved for administrators must now be extended across the enterprise. They must be applied wherever the action calls for it, and stay invisible where it does not. Your operating model must run at the speed of the system it’s protecting.
That is what Idira is built to deliver.
Learn more about next-generation identity security at paloaltonetworks.com/idira.
Next up, we’ll show you how Idira helps:
- Discover every identity, entitlement, and access path
- Control privilege at the moment of action
- Govern every decision and outcome
- Power AI-driven security for the agentic era.
FAQs
What is democratized privilege management?
Democratized privilege management extends the rigor of privileged access management beyond traditional administrators to every identity that carries privilege, including workforce users, workloads, machines, and AI agents.
How does democratized privilege management apply to non-human identities?
It brings continuous discovery, action-based privilege controls, and governance to machines, workloads, and AI agents. Access can be scoped to the work being performed and withdrawn when that work is complete.
Why does AI change privilege management?
AI assistants and autonomous agents can act at machine speed and may operate with delegated human or machine access. Identity security needs to discover those access paths, control privilege at the moment of action, and govern what happens next.