Skip to content
  • Home
  • Videos
  • Your Identity Stack from the Attackers POV • SHIFT On-Demand 2026

ON-DEMAND SESSION

Your Identity Stack from the Attackers POV: Rethinking Identity Resilience in the AI Era

Learn how attackers exploit ordinary identities, hidden group relationships, and routine changes to gain control—then see how auditing, rollback, and recovery help restore trust.

Video thumbnail

Key Takeaways

Expose Hidden Privilege

Trace relationships that reveal indirect administrative control.

 

Correlate Hybrid Events

Connect routine events across Okta, Entra ID, and Active Directory.

 

Roll Back Risky Changes

Reverse unauthorized changes before attackers expand their control.

 

Recover Trusted Identity

Restore identity systems to a point before compromise.

 

Reduce Attacker Dwell

Act quickly to limit persistence and operational impact.

 

Follow One Identity

Track one person across the full identity landscape.

About This Session

Identity attacks can look like normal employee activity. Identity attacks can look quiet because adversaries often log in as regular employees instead of breaking through a firewall. The session frames identity as the way into an organization and examines how attackers use normal access, hidden relationships, and routine administrative events to move toward control.

Five Questions Matter

The session walks through five questions: who controls the environment, whether one identity can be followed across the identity landscape, whether risky changes can be recognized, whether they can be reversed quickly, and whether trust can be recovered. The fictional Alex scenario makes each question concrete.

A regular user can hide a privileged attack path. Alex is an ordinary marketing employee, not a domain administrator. A phishing link gives an attacker access to his account, which leads through nested group membership to password-reset rights for a service account in Domain Admins. The scenario shows how preexisting relationships create an attack path.

Build Identity Resilience

The session closes with an identity resilience approach: hybrid auditing to correlate events, attack surface intelligence to identify control relationships, rollback to reverse risky changes, and full recovery to restore a trusted point in time. It returns to the five questions as an evaluation framework for identity and security teams.

See the full story across identity platforms. Hybrid auditing correlates events into one timeline so teams can identify risky sequences and act before attackers persist.

Restore Trust

Attack surface intelligence, rollback, and full recovery help teams reduce risk and reestablish trusted identity services.

Related Resources

WHITEPAPER

Build Identity Resilience

Learn how to protect, detect, and recover identity infrastructure.

Read whitepaper about Build Identity Resilience
SOLUTION BRIEF

Protect Active Directory

Explore visibility, control, and recovery for identity systems.

Read the brief about Protect Active Directory
DATASHEET

Identity Resilience for Okta

Protect Okta objects, view changes, and restore services.

Read datasheet about Identity Resilience for Okta

Frequently Asked Questions

How do attackers use valid identities to avoid detection?

Attackers can sign in with stolen credentials and appear to be legitimate users. Individual actions may look routine, so teams need correlated activity across identity systems to recognize the larger attack pattern.

Why are hidden identity relationships a security risk?

Nested groups and inherited permissions can give ordinary users indirect control over privileged accounts. Attackers can follow these overlooked relationships to reset credentials, escalate access, and create persistent administrative accounts.

How can AI accelerate an identity attack?

AI can help attackers research targets, discover permissions, prioritize useful attack paths, and identify ways to persist. It does not create the weakness, but it can make exploiting existing access relationships much faster.

Why is disabling a compromised account not enough?

An attacker may already have reset another password, created a new account, or granted administrative access. Disabling the first account does not reverse those changes or restore trust in the identity environment.

What capabilities help build identity resilience?

Hybrid auditing connects events across platforms, attack surface intelligence reveals risky control relationships, rollback reverses harmful changes, and full recovery restores identity systems to a trusted point before compromise.

Transcript

View Full Transcript

Hi and welcome to your identity stack from the attacker’s point of view. My name is Ron Robbins, and I’m a Senior Product Manager here at Commvault in the Identity Resilience Group. And I’m really happy you chose to spend time with me today and learn about, what an attacker can do in your identity, environment.

I’m going to start off with, a little bit of a scenario.

What do you picture when you picture an identity attack? Do you picture, somebody breaking into your firewall, maybe exploiting, your operating system, installing malware? You know, any of the type of big malicious attacks that we hear about in the media? A lot of times those things don’t really occur in identity attacks. Identity attacks are a little bit more quieter.

So today we’re going to try and answer five questions as we go through a fictional scenario of an attack and look at it from the attacker’s point of view.

We’re going to try and answer the questions of who truly controls my environment? Who has control or administrative control of my identity environment? Can I follow one identity through the identity landscape and figure out what’s actually happening?

Can I recognize a change that occurs in my identity landscape and determine whether or not that change is risky? Can I quickly reverse changes that an attacker might make, and can I recover trust in the identity system? Should those changes require permanent restoration? At the end, I’m going to return to these five questions, and we’re going to talk about how you can use these to evaluate your identity team and your security team, and just become safer in your identity environment.

Let’s first talk about what the market looks like. The attack surface is moved. Identity is now the way into your organization. 55% of organizations say they’ve experienced some type of identity attack or breach in the last 12 months. That’s a pretty high number when you think about it.

Attacks are also faster. 89% of organizations say they’ve seen some type of AI enabled bad actor operation in their organization over the last year, 89%.

So all of these attacks are being accelerated, they’re being modified. They’re being changed and adapted at just an enormous speed. And so it puts you even more at risk. Attacks are also evolving. They’re not just seeing phishing, but they’re also seeing things like compromised browsers. We’ve seen, MFA fatigue where users are constantly bombarded with MFA requests and eventually just agree, just to get rid of the MFA request.

And we’re also seeing things like token hijacking, where there’s several, you know, examples of this in the media where sessions or tokens have been stolen from users after they’ve authenticated, so that they can reuse that session and gain access to the system. One great thing I’d like to note about this particular slide, or one quote I like pretty well on the bottom of this slide is that adversaries aren’t breaking in.

They’re actually logging in. So when you think about it, they’re not breaking down things like firewalls. They’re not accessing, you know, malware, they’re not deploying malware. They’re actually just signing in as a regular employee. So it’s not recognized right away as, risky activity. They’re just signing in like a regular employee. So as we go through this scenario, think about that.

Think about how this regular employee gains access.

So let’s talk about what attackers see and what defenders see in the environment. They definitely have different views. When you look at a defender’s view, their basic option or their basic goal is to maintain control over their environment and it’s a difficult job. There’s disparate identity platforms in this particular scenario. We have three identity platforms.

There’s various objects like users and groups and policies applications.

They’re all stored slightly differently in each identity platform and each identity platform has its own nuances of policies and rules and roles, and it’s very complex. It’s very hard to keep track of. It’s very hard to track what exists across all your identity landscape in these disparate systems. Sometimes there’s different teams involved. For each identity platform that’s out there.

We see different Okta teams in different ad teams, and they’re separate. Sometimes they don’t even correlate activities. The defender or the attacker has one goal as well. Their goal is to gain control over your identity system. What can they gain control over? They don’t have to worry about the complexity of your identity platform. What’s the one identity or the one object that they can control that allows them further assistance?

Your identity system. How do they gain access to data, to applications, to make modifications to your infrastructure, or to elevate their privileges? You really need to understand this perspective, to understand how to secure your environment, to understand what the attacker wants to do, and to understand your view into the environment and how you can simplify that and make it easier for you to control.

So the first question of any attacker is who should I compromise?

And that’s going to bring us into this scenario. I’m going to introduce you to Alex. Alex is just a regular employee. He works in the marketing department. And let’s not hold anything against any marketing employees from this scenario. Marketing is great. I love marketing, but I just happened to pick a guy that works in the marketing department for this particular scenario.

He uses email, he has access to business applications, he modifies documents, he creates press releases. He’s not a domain admin. He’s just a regular employee. Nobody thinks of him as somebody that has access, in a sensitive way, or that he’s a sensitive identity. What we need to do is find a way to attack Alex so we can use several different methods.

We can use phishing. We can send him a link. We can use social engineering. Maybe we use AI to determine what kind of things Alex is into. And we, somehow, can impersonate maybe an I.T worker or a friend or a coworker, and, and use that to gain access. Maybe we can find a session he’s logged into, in a remote web browser and be able to hijack that session and gain access that way.

But no, what we determined was the easiest way to gain access to Alex was through sending him a link. An email where he thinks he needs to edit a marketing document. Maybe it’s a press release, and he clicks on the link and is asked for his credentials to log in. That was the easiest way to attack Alex, and that’s what the attacker used in this particular scenario.

He didn’t start with an admin account. He didn’t look for the privileged access. He looked for a way into the identity system. And that way in was through. Alex is just a regular employee.

So now once he’s in, here’s where we get to the heart of the story. Here’s where the attacker tries to determine what he can do with Alex.

What can I control with Alex? So the first thing he does when he gets in is he looks at who am I?

Well, he realizes that Alex is just a marketing employee, right? It’s not very exciting. But then he looks and he sees that Alex belongs to the marketing operations group. He’s a member of that group, and that might not be very exciting either.

I just see press releases and documents. But then he realizes that the marketing operations group is nested into another group. In that group is special marketing operations is nested into the application support group, and that application support group has the power to reset passwords for various service accounts. This was something that was probably set up in the organization long ago.

Maybe there was a situation that occurred that, you know, they had to reset or maintain passwords that were service accounts on servers somewhere, and that application support group was granted that right. And it was never revoked. It was never removed or evaluated in the future. So maybe it’s been there for a long period of time. Well, that marketing group has that access now and then.

If you look at what groups or what, accounts, that particular application support group has, the access to reset the password to. You can see there’s several service accounts on the attacker sees that there’s a service account that is actually in the domain admins group that he can now reset the password for. So if you look at the entire chain of events, there’s a group that Alex is a member of.

That group has a nested membership into another group that allows him access to reset a password for an account that’s in the domain admin group. It’s great. I mean, he’s struck gold. Basically, he’s able to log in as that service account and he’s able to create another admin account and add it to the domain admins group and gain persistent access as he goes through and moves through the identity environment.

Nobody, you know, thought Alex was linked to the domain admins group. This is totally hidden path for the attacker. But now Alex has control over the credentials of an account that belongs to a group that has admin access, and that’s what gave the attacker access to create his own account that has domain access. So there’s just several different events that occur, different relationships that occur.

And those already existed prior to the attacker joining, the identity system.

So when we look at this, when we look at, again, the defender view compared to the attacker view, we can see that Alex is just a regular person. And we as defenders or we as, people on the IT security team,
we look at our accounts and we see the admin accounts, we see the service accounts, we see what’s privileged in our environment.

We know that there’s certain accounts that are dangerous. We know that domain admins exist. We know that there’s other administrator accounts that exist, and that we have certain privileged groups that exist in our organization. Alex wasn’t in any one of those. If you went to the domain admins group and you asked, you know, did a query, is Alex part of domain admins?

He wouldn’t be part of domain admins. He wasn’t part of any privileged group or privileged access group. And so us just looking at the privilege accounts didn’t tell the whole story. It’s the relationships behind the scene. This is why we have something called tier zero awareness. You know, we have these tier zero objects that have direct administrative control over identity environment, but it should expand beyond that.

It needs to expand beyond just the tier zero accounts or the tier zero objects. You need to be able to see the relationships between all those objects to determine what truly affects your identity system, and what attacker can use to gain administrative control. Privilege not only belongs to like a privileged group, it’s also what you can control with that particular
privileged group.

So let’s complicate things even further. Alex is one person, right? Alex is one person that the attacker was able to, infiltrate and able to use their account. But Alex is represented in many different ways across the identity landscape.

You know, we have Okta in the organization for authentication. Alex might log in with Okta. He might use enter ID to access Office 365, and he might also use Active Directory to gain access to like legacy applications that, are on prem.

These are all different identity platforms. They’re different identity providers in Alex’s. In all of them. He might be represented, in a slightly different way, but he’s the same person. Once I basically infiltrate Alex. Once I have access to his account, I could gain access across the entire identity landscape and make a change that I’m able to do in any of those identity systems.

So there’s just basically one opportunity that the attacker needs to gain access to every identity platform. In your environment.

And this
kind of gives another picture to the complicated, task that we as defenders have, in our organization. You know, all of these what we would call routine changes or routine events that occur across our identity systems are all things that are just basically normal, everyday things that happen in our identity platforms.

And if we don’t have, a broad view of all of these events across our identity platform, we’re not going to be able to put together a story of how an attack is occurring. Like, let’s take this scenario, for example. You know, first, Alex logs in, has a successful log in with Okta, and it might be from a new IP address, but that necessarily isn’t a risky action, right?

Because users travel all the time. They travel for meetings, they go to events, they go to, conferences, you know, things like that. And, their IP address changes. So it’s not necessarily a risky activity. Plus MFA, the MFA challenge was approved. So I mean, it’s even less of a risky event because MFA was approved by Alex.

And and we know that he’s a real person there. And so he creates a new session. He accesses, you know, sales app data. Maybe that’s, you know, through a central ID account. And he’s, basically, does something where he looks up his group membership again, that that could be a routine thing. People look at their group membership from time to time.

So it’s not necessarily a dangerous thing. You know, he discovers that he’s in a nested group. And then he moves into Active Directory and he goes in any access is that, an account and resets the password for that service account. And then he uses that service account to log into a workstation and create another account and gives it admin access.

Now we’re starting to get to the risky behavior. But even those changes, even those events, are things that IT admins can do every day. People create accounts, new accounts every day. People might modify the admin group every day. People might, log in with other accounts on other machines. And so some of this stuff just appears as regular accounts.

It’s not until we correlate the timeline of events across all of the identity platforms that we see a complete story, and we see the risky actions, and we can put the pieces of the puzzle together to figure out that we’re actually under attack and that things are being modified and privileges being escalated. So it’s very important to correlate these activities.

Once the attacker’s in, he’s able to access directory data, he’s able to access sales data. He can modify group policy, and, you know, possibly deploy malware. So all of those things, that occur as a result of the infiltration, and you can see this entire story if you correlate these events across the identity platforms.

So let’s zoom in on just one event.

So that one event, that password change event is stored in. And this way you can actually see the ad log here on the screen. And you can see what was changed. You can see that it was just a a simple routine change. An event that looked very routine.

They just logged in with their account. They reset a password, for a particular account.

It’s there’s no reason to raise the alarm bells. That happens every day. People get locked out of their accounts. Service accounts might need to be reset or, credentials rotated. You can see actually who performed the event. And nothing about this event appears dangerous. It doesn’t appear risky. It’s not until we correlate this particular event with the other events on the previous screen that we can actually tell the full story, that we’re under attack, and that there’s actually risky behavior, happening in the organization.

So that simple password reset caused a much bigger impact to the identity platform before we just had, you know, Alex, a regular employee, right. And, Alex, all these changes were in place for Alex prior to the attack beginning, Alex was already a member of the operations group. That operations group was already nested in the sales application support group.

And those, service accounts already existed. And the access that those service accounts had already existed,

all the attacker had to do was simply perform a password reset, and he gained access to a new account that allowed him to create his own account, to persist in the environment and to add additional control for himself over the environment. Now imagine for a second that your security operations team, they actually see all these events occurring, and they see Alex log in, they see Alex reset passwords, they see Alex create a new account, or they see that service account, create a new account.

And they there they realize, oh, Alex has been compromised, I’m going to go ahead and I’m going to shut down Alex’s account and I’m going to make him reset the password for that account or or just lock his account out. It doesn’t matter anymore, because after the attacker has already used Alex’s account to get into the system, they really don’t need the account anymore.

They’ve already created a persistent account for themselves, already gave themselves domain admin access, and so they really don’t need Alex anymore. And just shutting down Alex’s account does not solve any problems. It does not restore trust in the identity system. It does not roll back changes that provided the attacker access into the system.

Let’s talk about what I can do in these scenarios.

Up to this point, every single thing that I’ve talked about can be done without I. Let’s just be honest with that. You don’t really need AI to do any of these things. We can send out phishing attacks. We can gain control over an account. We can elevate our privilege within that system. I just accelerates the entire process.

Right. I will help us determine who to target. Which identities should we compromise? And even before we compromise the identity, it might go out and do research about Alex and figure out what kind of things are Alex and what kind of things are Alex? Is Alex into, what relationships does he have with others? Is there a way to, you know, use some of those relationships or his position in the organization to create a social engineering attack to attack?

Alex. It might make the attacks look much more intelligent and give us a faster way to attack Alex than we ever had before. Then there’s the discovery. Once we’re in, once we’ve infiltrated Alex’s account, we can go in with AI, and we could do a complete discovery of what Alex has access to, what groups he’s a member of, and what groups those have rights to, or what those groups have rights to do in the organization.

In this case, reset passwords right. And then it helps us prioritize the attack, helps us prioritize exactly what we should focus in on. In this case, we can use AI to determine that that one service account has domain admins. We might not have seen that otherwise. We might have just gotten lucky and clicked on one of the service accounts of maybe the hundreds that the application support group had access to.

And and just gotten lucky to see that it had domain admin access. With AI, we can do a much broader search and much broader analysis and do it extremely quickly to determine out what to determine what we need to prioritize as our point of attack. And then we can have, I suggest to us, how do we expand and persist in this particular identity environment.

In this case, it was to create a brand new account and to add it to the domain admins group. It might be different in some other cases. And then also we can use AI to determine what is the best impact that we can have in this organization. How can I access data? How can I use it to my advantage?

How can I create objectives for myself that would give me the most profit, or give me the most impact to the organization that I want to create as an attacker? One thing that’s really important to note I did not create any of the weaknesses in the environment. I did not create any attack paths in the environment. All I did was help the attacker quickly exploit one account, and then quickly determine what that one account could do within the organization to exploit an attack path that was really already there, that the organization has created for themselves.

So now it brings us to identity, resilience. So all this stuff that occurred in this particular scenario, if we had an identity resilience in place, I think that there’s some things that we could do to help protect ourselves and recover from these attacks. You know, should they occur.

The first one is hybrid auditing. We need to be able to see the full story across our entire identity landscape.

And so being able to correlate and collect all those events that occurred that looked like disparate routine events in each identity platform, being able to correlate that and put it into one identity timeline really tells the whole story of the attack, and we could really see the behavior as it plays out across our identity platform and be able to take action, as it occurs.

We can have it flag for us what the high risk events are that occur in our identity platform, and we can make sure that we can recover from those events. And we can also provide this unified timeline. But the timeline isn’t only just about time, it’s about what objects were affected, what users performed those actions and when those actions occurred.

And to be able to see the complete story and know where the point of recovery is at, for our particular system. And then we need attack surface intelligence. Attack surface intelligence is going to help us identify the objects that have administrative control over our environment. And not only that, it’s also going to identify the relationships between those objects and other objects that might allow an attacker to gain administrative control.

If a change expands the footprint of the attack surface, it’s going to tell us that as well. It’s going to be able to help us determine, exactly what changes took place that allowed more objects to have administrative control in our environment, and then it’s going to prioritize the risky changes and allow us to figure out which ones to remediate, to immediately roll back the changes.

And then of course, we need rollback. We need to be able to quickly revert changes back to what they were before, prior to the attacker getting in. So we be able to determine which changes are risky, take the remediation advice and roll back the changes that it suggests that we roll back in the environment that are risky or affected our systems or increased administrative control.

We need to be able to reduce the attacker dwell time within our environment. So the quicker we can rollback changes, the better we can defend against that. And then should an attacker break through, should an attacker elevate their privilege, should they be able to create additional accounts and persist in your environment and make a ton of changes that you just didn’t necessarily know about or authorize?

We need a way to recover and restore trust in the identity platform. That’s where full recovery comes into place. We need to be able to validate and reestablish trust. That’s the most important thing to be able to, determine what’s changed and actually restore back to the point in time before those changes occurred so that we can, basically reestablish trust in the identity platform.

It’s very important prevention matters, right? It’s important to be able to detect all these things with hybrid auditing, to be able to determine what the risk is with attack surface intelligence, rollback those changes immediately by, making sure that we don’t allow the attacker to dwell in our environment and then perform a full recovery if we need to restore trust.

All those things are important.

So now that we’ve covered that, let’s go back to those five questions I talked about at the beginning of the presentation. I think that you can take these with you and use these to help evaluate your current identity, security and where it’s at today.

So go back and Azure Identity security teams these questions who can truly controls my environment?

Can I quickly determine who controls my environment. What are the objects that have administrative control in my environment, and what are the relationship? Those those objects have, that allow them administrative control in my environment. Also, can we follow one identity across the identity landscape? Most organizations, and I think you would agree with me, most organizations have many identity providers.

Can I follow that one identity and how it’s represented, how it’s represented in all of the identity platforms, and see the changes that occur for that identity or the changes that identity makes throughout your organization or throughout your identity landscape. Can we recognize when a change expands control? There’s a change to a service account, a change to a group, that all of a sudden elevates that group or that service account to have admin privileges.

Can I recognize that change right away and see it? Can I quickly reverse those changes? So if I see the story of an attack unfolding and I see all the activity or all the, events that occur with that particular account, can I roll back those events very quickly to undo changes that attackers making in real time? And then if I don’t catch the attack, can I roll the system back or can I recover, my complete identity system to reestablish trust.

So can I find a point in time before the attacker got into the system and completely recover it and establish trust in the identity system again, to completely recover it? You know, one thing that, I think is an important lesson here that we should take away from this is that the attacker didn’t attack, domain that had been account.

The attacker didn’t look for, privileged accounts to attack. He looked for one user that was vulnerable, and that user was an ordinary, everyday user. And he used that user to gain privilege and use that user’s relationships to gain access into the environment. So it’s important again to understand the control, recognize what changes in your environment to determine where the risk is, and then take your environment back, recover it, and roll back those changes that occur in your environment.

I want to thank you so much for joining me today. I want to thank you for the opportunity to be able to talk to you about identity, resilience, and also talk to you about this particular scenario that I think it’s a common scenario, that could occur in most, organizations. I hope that you can take back with you some of those action items and evaluate, your identity, security and your organization and evaluate how you might recover from an identity event that occurs in your organization.

Thank you so much for your time today. And thanks for joining me.