Agentic AI Security: How to Protect Autonomous Workflows by Palo Alto Networks

Note to self, never follow Nikesh on stage again. Well, good morning everyone. First and foremost to all our customers, partners and all my colleagues from CyberArk and Palo Alto Networks. I'm absolutely delighted to be here today. It is day zero in the world of cybersecurity for me. But I'm thrilled to be here and I look forward to working with all of you in the coming future. Thank you very much for that warm welcome.

This is an important moment and I'm excited to be here for that. What we are announcing today is much more than the launch of a brand. The launch of IDRA is very important because for us it means a new platform direction. And the reason why this matters is really very simple. Nikesh talked a lot about it. The technology industry overall has been very prolific, very, very innovative. But one thing that you can hold us accountable for is that we've been building new features, building new capabilities. We do what we do best and then we toss it over the wall to our customers and we leave the challenge of making all of that work in your environment, in your architecture to you.

This means there's a lot of disparate data sources. There are disconnected workflows. There's a lot of manual integration. And this reduces the opportunity to accelerate time to value. All of this is happening while you're dealing with a threat landscape that keeps on evolving. A phrase you're likely to hear over and over and over again is that attackers are not breaking in, they're actually logging on. And this is what we are seeing, more and more attacks start with identity. They start with maybe, you know, unauthorized privilege. They may start with credentials that are outdated. They start with access that looks legitimate until it's abused and it's too late.

So this old model where privileged access is treated like a narrow, isolated problem is just not enough anymore. All of you, our customers, need a much more modern approach and that is what IDRA is all about. It's about bringing privileged access and identity security into a broader platform model. It's about reducing the burden of integration for you. It's about delivering that value of integration as a built-in capability, not bolted on after the fact.

When we think about platforms, IDRA is actually a critical third pillar to the broader Palo Alto Networks portfolio of capabilities. It sits besides network security which is Strata, cloud security which is Cortex, and now identity security, all working closely together with AI. This is important because it actually tells you how central identity and privilege have become to the overall security architecture. It's no longer a side category. It's fundamental and it's foundational. And this is where the platform advantage matters so much. It means for you less complexity. It means more context and it means better outcomes and an accelerated time to value. At the end of the day, you as our customers should not have to take on this integration challenge, making all of these piece parts work together so you have a comprehensive capability for managing the threat landscape. It's our job to make that easier for you. It is our job to deliver a platform that works together by design. That is the promise of IDRA. A modern platform for privileged access and identity security. Built for the way attacks happen now and built for the way that customers need to operate going forward.

Hello Impact everyone. It's really, really incredible to be here with you today. We are standing in Austin, in the heart of Austin, also known as the Silicon Hills. A place that has spent decades in redefining what it means to be a global hub of innovation. This city doesn't just build technology, it reimagines it. Austin was the first place in the world to see the PC industry completely disrupted and it continues to lead the way in advanced manufacturing and infrastructure for the commercial use of AI. This is the place that chooses to scale rather than just experiment and it makes it exactly my kind of place. It is an honor for me to stand here with you today at the dawn of a new era for identity.

And to understand where we are going, we have to look back on how we got here. Over 25 years ago, CyberArk built a safe. It was designed with a simple and noble purpose to protect digital assets. But then something interesting happened. A customer told us they were not just using the safe for general files. They were using it to store admin credentials. They also told us they didn't really like the word safe because it felt really just a bit too small. They wanted to call it a vault. That was the spark. That was the moment privileged access management market was born. And at the time the analyst at Gartner gave it a clinical name SUPM, stands for super user privilege management. And for the longest time that exactly what PAM was. It was super user privilege management. It was a world of physical servers, local admins and a handful of people who held the keys to the kingdom. It was a very simpler time with a really well-defined perimeter and a manageable number of targets.

Those days are gone. Everything has changed. Our environment, our identities and the entire attack landscape have shifted fundamentally. We have moved from a single isolated data center into incredibly complex environments. We are managing distributed clouds, hundreds, hundreds of SaaS applications and thousands of microservices. This has led us to an ever-increasing attack surface. Every point of connection is now a potential point of entry. And to make matters worse, we are no longer facing the opportunistic hacker of the past. We are facing sophisticated and AI-armed cyber criminals. These adversaries using the same identity tools we use to automate their defense, making identity the number one target. Identities, privilege and risk are already pervasive throughout your entire environment.

Let us start with the human users. We used to think privilege was reserved only for the IT admins in our environment. Today, every human identity can become a privileged identity. Whether it is a developer pushing code, a cloud operator managing infrastructure, or a business user accessing sensitive data, they all have the power to impact the business. The data tells us a dark story. 96% of human users today have access beyond what is strictly required to do their job. 96%. We are living in a permanent state of over-privilege. This uncontrolled privilege means that for almost every person in your organization, an attacker does not need to find a vulnerability. They just need to find the login.

But humans are no longer the primary resident of the network. Machines have taken over at a scale that is really difficult to comprehend. For every human being in your enterprise, there are now 109 machine identities. These are the non-human identities like API keys, service accounts, and secrets buried in your CI/CD pipelines that run your business at machine speed. This is hidden access to your highest risk systems and data. They are the silent majority of your identity state. They are growing exponentially and in most organizations they are operating entirely in disconnected silos.

Now let's add the newest and most volatile persona to the mix which is of course the AI agents. These agents have memory, they have skills and they act autonomously. They are being deployed faster than any technology in history. Yet governance is non-existent. 94% of companies admit they lack the proper security controls for AI agents. You know, we have allowed AI adoption to drastically outpace our controls. These agents are being delegated to act as us on behalf of but they operate in a governance vacuum where we cannot really reduce their credentials. We have given AI the arms and legs to act but we have not given it the nervous system of security.

So why are we in this position? We're in this position because the industry is fighting a platform-scale war with a collection of fragmented point products. The issue is in how the identity market itself evolved. Standard access management focused on operational management not security. Traditional PAM solutions focus narrowly on IT admins. Legacy identity governance is pretty complex and slow. Machine identity grew in a disconnected silo and AI identity is emerging as yet another layer. Sometimes they come from different vendors. Sometimes they come from the same vendor but remain loosely integrated. They always leave gaps and to compensate, organizations keep adding more and more tools like identity threat detection and response. These silos create the gaps where attackers live. When your identity tools do not talk to each other, you cannot see an attacker move from hijacking a human session to an unmanaged machine to an AI agent. Silos do not just create complexity, they create catastrophe.

And the result of this fragmentation is inevitable. Identity is the number one attack vector today. Nine out of 10 organizations experience at least one successful identity-related breach. And Palo Alto Networks Unit 42 tells us that in 90% of the incidents an identity weakness played a role. Attackers are no longer breaking in. They are logging in and they steal credentials, abuse standing access, escalate privilege and move laterally using legitimate pathways. When identity is compromised, everything downstream, the cloud infrastructure, SaaS platform, the data itself can become exposed. Identity is no longer just part of security architecture. It is the attack surface in the AI enterprise of tomorrow. This will become even more challenging.

The enterprise is being redefined not by the software alone but by how work gets done. From employees using identity browsers to sync CRM data to developers building AI-driven endpoints, AI is becoming the digital surface of the enterprise. We are moving toward a world where identity will become more dynamic, more distributed and much more chaotic. Risk is turning exponential. AI agent growth is now significantly outpacing human growth. We are no longer just managing people, but a high-speed autonomous workforce. This creates risk we have never seen before. Unchecked agent access and weaponized intelligence moving at machine speed cause traditional defense to fail. This compounds daily into a state of identity chaos. We have reached the point where the speed of the AI agents has officially outrun the manual speed of the administrator.

You heard Nikesh. We have reached our inflection point. The state of the identity market must evolve fundamentally in order to keep up with the agentic future we have described today. We can no longer pretend that adding another tool, another silo capability will fix the problem. Yesterday's silo simply cannot solve the problem of tomorrow. You as an identity and security leader must meet this moment. You are the architects of the next era of defense and you lead the transition from administrative management into active security.

What is required is a cohesive identity security platform. The only effective answer to the existing chaos is a single unified architecture that protects all identities including human, machine and agentic. The modern enterprise requires breaking down the silos and deep inefficiencies found within the current point product reality. This means redefining security controls that were once reserved only for a few users and establishing them as a universal standard for all. Such an approach fundamentally disrupts the legacy market of PAM and IGA. Those markets were designed for a static world. But in the agentic era, it demands a platform built for a dynamic one. This is a mandatory evolution of identity security. Bringing together world-class identity intelligence with an integrated security fabric to create a foundation that cannot be broken.

This is why we are here today. I'm incredibly proud to introduce IDRA. IDRA is a cohesive identity security platform built especially for the AI enterprise. We have taken the trusted and best-in-class DNA of CyberArk and infused it with the global scale and AI precision of Palo Alto Networks. IDRA is a cohesive system that can continuously discover all identities, control privileges and govern access across the entire enterprise. It enables organizations to both find and remediate identity risk at machine speed. Discover, control, govern. These are not new ideas, but the way we apply them has fundamentally changed.

Let's start with discovery. For a long time, discovery meant inventory, a point-in-time view of what exists in the system. But that is no longer enough because identities do not just exist. They evolve. They gain access. They lose access. They interact with new systems. They behave differently over time. So discovery has to be continuous. It has to be contextual and it has to be intelligent. This means not just identifying the identities but understanding how they behave, what they access and where they introduce risk across environments. And it means surfacing that risk in a way that teams can actually use and with context to understand it and a remediation path to reduce it. This is why discovery in IDRA is paired with enhanced context, with centralized visibility and AI-driven risk analysis, not just a list of identities.

Once you've figured out and you have the level of visibility, you can move into control. And this is where one of the biggest shifts happens because historically access has been static. Users were granted permissions. Those permissions persisted and over time those permissions accumulated. That created standing risk because once entitlement exists, it can be exploited. What we are doing instead is making privilege dynamic. Access is granted when it is needed. It is evaluated continuously and it is removed when it is no longer required. So instead of asking who has access, we start asking a more important question. Who should have access right now? And that model applies across the full identity spectrum. For human identities, that means zero standing privilege and just-in-time access to modern systems. For machine identities, it means stronger workload identities and secrets control. And for agents it means fine-grained action control and governed delegated access.

And then governance. Governance ensures that this remains true over time. Because identity security is not just about the moment of access. It's really about the full lifecycle. How identities are created, how access is assigned, how it changes over time and how it is eventually removed. That lifecycle has to be continuous. It has to be automated and it has to be consistent because the risk does not come from what you configure once. It comes from what drifts over time. And the key shift here is that governance is no longer a periodic compliance exercise. It becomes a continuous operational control. AI-powered policy helps move organizations from manual reviews and siloed workflows to automated governance to keep pace with the scale and velocity of the modern enterprise.

Our vision is clear. One platform, every identity. Today you will see IDRA in action across human users, machine identities and AI agents. We are ending the era of identity chaos.

Well, thank you. It's amazing to be here in Austin and to see so many familiar faces. Thank you for spending time with us. Now, I want to start with something that on the surface feels very familiar, but when you really look at it, it has completely changed. And I'm talking about privilege. For a long time, we have thought about privilege in a very specific, very narrow way. In fact, privilege belonged to a small group of people, IT administrators, who held the keys to the kingdom. That was the original problem that PAM was built to solve. And CyberArk solved it better than anyone else. But here is the reality we have to face. That world no longer exists. In every modern enterprise, every single user can become privileged based on the data, the SaaS applications they can access. Privilege has been democratized, but not in a way that makes us safer. Today, your developers are pushing code directly into production. Your cloud operations, well, they are managing infrastructure from a single console. Your data scientists are running models against your company's most sensitive and proprietary customer data. And now business users just like me, we are living inside SaaS applications that hold the crown jewels of your information. These are all high-risk identities that can impact entire systems with a single keystroke. Yet, despite this reality, only 22% of organizations attribute privilege to every human identity. This is a massive disconnect. It means that for 78% of organizations, true identity security is only being applied to a tiny subset of users. Human privilege is pervasive. It is uncontrolled and it is an open door for attackers. This disconnect creates what we call the privilege exposure gap where privilege lives today and where security is actually applied. Now there are three primary reasons why this gap continues to widen. First, we are suffering from fragmented visibility. Traditional tools were built for a static world. They provide limited insight across modern workloads and cloud-native identities. This is what we call identity debt. Hidden in plain sight, accumulating interest every day you don't address it. Second, we have a problem with persistent risk. We are still relying on static, always-on privileges. In a modern environment, standing privileges should be the exception. Yet somehow, they've become the default. This creates a permanent attack surface, allowing adversaries to exploit that standing access whenever they choose. And finally, there's complexity in our tools. Traditional systems require heavy manual deployments that can slow down your teams. This forces business leaders to make an impossible choice between security and velocity. When you combine these factors, over-privilege and dormant access accumulate rapidly. The attacker can use these trusted paths without triggering alerts. Risk and cost compound over time, and this gap is only growing larger. So we have to be honest with ourselves as an industry. The existing model is broken. It simply cannot keep up with pervasive privileges and the sheer velocity of AI growth. This moment requires more than just an update. It requires a fundamental category shift. We have to redefine how we deliver privilege security across the entire enterprise. We cannot just iterate. For too long, the industry has accepted a two-tier system. High-grade security for the few and basic authentication for the many. But attackers don't care about our internal definitions. They care about access. Meeting this moment means moving beyond the silos of the past to build a system that treats every identity as a privileged identity by default. We must be willing to disrupt ourselves to meet the needs of the AI enterprise. And it is with that spirit that I am incredibly proud to introduce our next generation of human identity security. IDRA modern privilege access management. IDRA sets a new standard for PAM. By eliminating standing privileges and extending dynamic privilege controls to all users, we are democratizing privilege controls. And this will help ensure that all identities are secured, not just managed. We are building this on three core pillars. Full discovery to eliminate blind spots. Dynamic privilege access to extend controls to every identity. And modern PAM and radical simplicity to make security a force multiplier. Now, let's take a look at this first pillar of discovery. But this is not discovery as we have known it. Not a static inventory, not a one-time scan. This is continuous AI-driven discovery. Understanding where privilege exists across the enterprise, across users, across cloud environments, across SaaS applications, but more importantly, understanding context. How these identities behave, what they access, and where they introduce risk. We are introducing the remediation agent where visibility finally meets automated action. The goal is to replace your blind spots with continuous visibility and AI-driven analytics that surface and remediate identity risk across the entire enterprise. This agent is a true game-changer. It analyzes your specific tenant account onboarding patterns to predict and recommend the best controls for every discovered account. This is the find it, fix it capability the industry has been waiting for. The remediation agent acts as a force multiplier for your team. It can reduce your onboarding cycle time by up to 66%. We're talking about moving from a 21-day manual fragmented process down to just seven days. And we do this all with a human-in-the-loop philosophy. Every AI recommendation includes a confidence score and the specific onboarding reasoning so your team remains in control while moving at light speed. Now, to show you exactly how we turn visibility into action, let's look at how IDRA's AI remediation agent tackles the massive scale of unmanaged accounts across your enterprise. Every organization is buried under identity debt. For a security admin, the IDRA risk management dashboard maps that debt, revealing a high-risk baseline of unmanaged admins and exposed secrets that grow faster than we can secure them. We can't fix it all at once, so we prioritize. By identifying our top five hotspots, we see the structural risks that matter most, like orphaned users with privileged access and compromised shared app credentials. But we don't stop at visibility. We use Kora AI to turn these insights into a plan. We simply ask Kora to secure our most privileged accounts, providing instant visibility into the exposure, and remediation is now automated for built-in and service accounts. The AI agent does the heavy lifting, analyzing context and onboarding accounts in bulk. This is PAM at the speed of business. For personal privileged accounts, we remove the risk entirely. We move users from permanent standing access to zero standing privilege. So access only exists when the work is actually happening. The result, we've shifted the needle. Our risk score is trending down because we've neutralized the exposures that matter most. Moving from identity debt to a hardened posture. Finally, we ensure that debt doesn't crawl back. Continuous governance keeps the environment secure long after the initial cleanup. What you just saw was the shift from manual discovery to automated remediation, empowering you to not only find risk, but to fix it. Now, I love hearing from our customers and how the innovation we're driving is helping them every day. So, let's take a listen to Brian.

Identity is, I really think, the place to be, and you've got to get that right to enable the business. How do we change that paradigm where we can reliably leverage AI to solve our problems at speed, much, much faster? And again, some of the remediation solutions are some of the things I've been looking for. We're sort of at a balancing point. Too fast and we're not going to think things through all the way. Too slow and the whole industry and the whole capability set that we're trying to secure is going to get away from us. We need reliable, repeatable solutions that are at speed. I think more and more high-quality platform solutions are really important. Palo Alto has a lot of best of breed that they bring to the table. It's the same innovation we've counted on for many years going forward as part of a very, very state-of-the-art Palo Alto Networks family.

We move on to controls and this is where the biggest shift happens. We move away from where identity is static and the idea that access should exist permanently and towards a model to have access exist only when it's needed. This is dynamic privilege and it's delivered through a combination of models. Zero standing privilege, just-in-time access, and contextual enforcement. Now, zero standing privilege means exactly what it sounds like. There are no permanent permissions, no standing access, nothing exists unless it's required. Even if a compromise happens and the credential is taken over by an attacker, there is no entitlement for them to use and leverage laterally. And that changes everything because if a permanent entitlement does not exist, it cannot be exploited. For scenarios where access is required temporarily, we use just-in-time access. Permissions are granted for a specific task for a specific duration and then removed immediately after. And the key here is updating and extending this model so that it's not just for traditional environments. It's for modern environments. That includes zero standing privilege and access for managed Kubernetes and Entra ID groups helping customers bring dynamic privileges to systems that have become central to modern operations. It also means shifting PAM left with Terraform and CLI-based automations that allow customers to embed privilege controls into deployment workflows instead of layering them in after the fact. And privilege controls are not just for IT administrators anymore. We are extending true security to all human users. Whether it's a developer, a data scientist, a business user, we are providing seamless privilege session management for the web applications where your most sensitive data lives. We are shifting from a workforce state of static standing privilege to a secure dynamic state of just-in-time access. Now, to show you how this transformation impacts the daily experience of an employee, let's walk through a day in the life of a platform engineer operating in a world of zero standing privilege. Alex is a platform engineer and his day begins from a trusted foundation. His workstation secured by Cortex, IDRA Endpoint Privilege Manager and Prisma Secure Browser. If Cortex identifies a threat, it triggers EPM to impose a stricter policy posture to neutralize risk. Alex starts with zero standing privilege. Because of his role in the Project Phoenix production deployment, the platform has already paved the path for the infrastructure and applications he needs. This isn't manual policy creation. It's deployed as code via Terraform. Security is baked in the pipeline. This allows him to jump straight into the mission. First, he needs to push an updated Intune policy for devices that are part of the project. Alex gets scoped just-in-time access to Intune and his privileges expire at the end of the session. Next, Alex needs to monitor the health of Project Phoenix using ZSP for SaaS applications to access the Chronosphere dashboard. He identifies a latency alert on the production cluster. Alex has the necessary visibility to diagnose the root cause without standing privileges. To resolve the problem, he uses the IDSC CLI to gain native access to the production cluster to remediate it. All with zero standing privilege. Finally, the CISO's view. The command center is the high-definition proof of our zero trust strategy. We see the full story. Not just Alex's ZSP access, but the activity of our machine and AI identities as well. The Phoenix production rollout is a success. The audit trail is complete and the attack surface has been reduced by shifting from static permissions to dynamic session-based access. We've empowered Alex to move at the speed of business without leaving an open door for attackers. Now just as importantly as what we do is how we deliver it. With IDRA we have completely reimagined the user experience through a unified command center. It provides real-time insights from workforce users and vendors to service accounts and AI agents. Now it doesn't just show you identities. It shows you exactly how they connect to your systems, the privileges that they hold, and the risk they pose. But this is more than just a dashboard. It's an entirely new way of working. By implementing a unified policy model with guided outcome-driven workflows, we've fundamentally reduced the overhead of operating PAM at scale through self-service deployment. From the command center, your security teams can finally monitor identity posture in real time, enforce policy, and investigate incidents from a single unified console. The PAM you know and love is finally ready for the future with IDRA. Everything we've discussed today, from discovery, to dynamic access, to radical simplicity, it all leads to a very different outcome for your organization. First and foremost, you eliminate standing access by ensuring that no identity holds access longer than necessary. You remove the permanent attack surface that adversaries rely on. You reduce lateral movement when sessions are contained, monitored, and bound by just-in-time controls. An attacker who gains a foothold has nowhere and nowhere to go. But perhaps most importantly, you can extend privilege controls across the entire workforce. We are moving beyond the era where high-grade security was reserved for a small elite. Every single identity has access that matters because as we have seen, that is exactly where the risk lives. When you address that risk in a way that is dynamic, contextual, and simple to operate, you do more than just add a new tool. You fundamentally change the security posture of the entire organization. You move from a reactive state of identity debt to a proactive state of active security. IDRA Modern PAM enables us to close the exposure gap. This is the end of the always-on era and the beginning of a resilient AI enterprise. Now, to show you how these same privileges apply to the exploding world of non-human identities, I want to welcome to the stage Avahai.

Woo! Thank you, Amy. That was fantastic and perfectly sets up the other half of the story. Amy just showed us how privilege is expanding across the workforce. But when we look at machine identities, we're dealing with an entirely different scale. Look at the swarm on the screen. The ratio is already 109 machine to a single human. It's going to be 110 soon. And with the fast adoption of AI agents, it's growing faster than ever before. Think about what makes up your infrastructure today. Every automated process, every microservice, every container, every AI agent, these are all identities. They all need to be authenticated. Many of them need privileges and secrets to do their job. And every single one of them introduces risk. The wave that you see on the screen represents sprawl. Not just a sprawl of secrets, also a sprawl of unknown machine identities and scattered vaults. And together they create a massive hidden security blind spot. And the sheer number of this machine is exploding. They spin up, they do their job and they disappear. And because of that, the security solutions around them must evolve. Because when you have unknown identities and scattered vaults, you lack centralized governance and you lack control. If you see just a sea of alerts without any context, where do you start? How do you prioritize? The manual processes can't simply keep up with the modern technology and scale to secure dynamic environments. Automation is an absolute must because if our solutions do not evolve, we don't just lose visibility. We lose control. All right. So, we introduced the risk. Now, let's understand how we solve them. Previously, we introduced Secrets Hub to give you centralized visibility and control over your external vaults. But the environment is moving faster than ever. So today, we're adding the automation at scale that machine identity security truly needs. Whether your secrets are in AWS, Azure, GCP, HashiCorp, we find them. We remove the friction. No more complex setups. No more manual configurations. You can just onboard them to our platform with a single click and start managing it centrally. But here's the most important part. We've added all that power without changing the developer workflow. Your developers can stay on the environments they love. They keep their velocities. But you, all of you, you get centralized control that you've been looking for. We didn't just update Secrets Hub. We made it work at the scale and the speed of the machine. Now, let's take a deeper look. We're also adding a brand new view of risk and centralized visibility for machines. Managed secrets, unmanaged secrets, external vaults, even our own SaaS. Everything is now unified in a single automated view. But we don't stop there. We're going deeper. We're going into the ephemeral world. Those identities that live just for a second, we see them. All those pesky Jenkins CI/CD pipelines that used to be blind spots, we inventory them for you. We discover credentials and workloads automatically. We're taking the unknown out of your environment and replacing it with absolute clarity. Lastly, giving machines a strong way to fetch secrets has been our focus for years, but we want to take it a step forward because I showed you a sprawl, but how can you regain control over your sprawl without having a true zero trust solution for machines? Up until now, most of the industry secured this application based on what they know, a secret or a credential. But here's the problem. If an attacker steals a secret, it's instantly trusted. This is a fundamental identity problem. We have to shift the focus to who the app is. So what's our solution? Today we're introducing Secure Workload Access which solves this by issuing universal SPIFFE-based identities. It doesn't matter if it's a Kubernetes pod, a VM or an AI agent. Before they can do anything, they have to prove their DNA through continuous attestation. But here's the real breakthrough in all this. Those identities that we're issuing, they're ephemeral. They live for minutes, even seconds. And in case the identity is compromised by an attacker, somehow it is already gone. We are dramatically reducing the risk of forever secrets. This is the ultimate shift left. We manage the machine identity from day one. We're not just securing your workloads. We're giving them proven identity that finally makes zero trust for machines a reality. Now, we talked a lot. Let's see it in action and walk through how we are protecting machine identities.