ON-DEMAND SESSION
Secure Enterprise Data Activation for AI
Learn how to discover sensitive data, govern access, adapt controls, and restore trusted operations as enterprise AI use expands.
Key Takeaways
Know Where Data Lives
Discover and classify sensitive data across cloud, SaaS, endpoints, and on-premises environments.
Govern AI Access
Control which people, applications, and AI systems can use sensitive data.
Adapt Controls Quickly
Change access, masking, protection, and security policies as risk conditions evolve.
Protect Prompts and Responses
Detect sensitive information in AI interactions and apply real-time privacy controls.
Keep Humans in Control
Use contextual signals to guide decisions while authorized people approve recovery actions.
Restore Trusted Operations
Recover clean data and critical services to a trusted business state.
About This Session
Enterprise AI creates value only when organizations can activate data without losing visibility, control, or recovery readiness. This session explores how AI-driven attacks, expanding identities, and hybrid data sprawl reshape enterprise resilience. Learn how data discovery, classification, access governance, adaptive controls, and clean recovery work together to help organizations use sensitive data responsibly.
Know Your Data Risk
Security teams need context about where sensitive data lives, who or what can access it, which policies apply, and where exposure exists. This visibility helps teams prioritize risk and make faster, better-informed decisions.
Secure data activation requires visibility and enforcement across users, AI systems, structured data, unstructured data, and hybrid infrastructure. Data Posture 360 discovers and classifies sensitive data, while Data Access Manager applies real-time access controls, masking, approvals, and time-based access.
Adapt as Conditions Change
When credentials are compromised or lateral movement is detected, teams may need to restrict access, increase protection, mask sensitive data, or change policies without slowing the entire organization.
Recovery returns critical data and services to a trusted state when prevention and access controls are not enough. Context from security signals, identities, and data classification can support cleanroom investigation, quarantine, access changes, and recovery from a known-good copy.
Keep People in Control
Automation can coordinate information and proposed actions, but authorized people should review the context and approve consequential recovery or access decisions.
Related Resources
Data & AI Security
Discover sensitive data and govern access for people and AI systems.
AI-Ready Data
Prepare governed, compliant datasets for RAG and AI development.
AI Data Resilience
Protect enterprise AI data, models, and supported compute.
Frequently Asked Questions
What is secure enterprise data activation for AI?
It is the practice of making enterprise data usable by approved people and AI systems while maintaining discovery, classification, access governance, privacy controls, monitoring, and recovery readiness.
How does data discovery reduce AI risk?
Discovery shows where sensitive data lives and how it is exposed. Classification adds context so teams can apply access restrictions, masking, retention, remediation, and protection based on sensitivity and business value.
How can organizations govern AI access to sensitive data?
Organizations can apply policies based on the requesting user, application, or AI system, using controls such as dynamic masking, redaction, row-level security, approvals, and time-limited access.
Why must data security controls adapt quickly?
AI use, identities, permissions, and threats change rapidly. When risk conditions shift, teams may need to restrict access, mask data, increase protection, or change policies before exposure expands.
What role does recovery play in secure data activation?
Recovery provides a path back to clean data and trusted operations when prevention fails. Security context can help teams investigate in isolation, quarantine risk, revoke access, and restore known-good copies.
Transcript
View full transcript
Hi everybody, and welcome to our session Secure Enterprise Data Activation for AI. I’m Ben Hertzberg. I have Yoav Cohen and Jose Gomez with me. And let’s kick it off.
So I think that we all know where we are standing right now. AI is here. It’s part of the enterprise. Productivity cycle. From a recent study, 93% of CFOs are using, AI for meaningful, operations, and 90% of AI governance executives say that data risks are material.
So basically, we are all addicted to AI, using it at work for many different things, including, key initiatives. And we all know that AI is also risky. But obviously we can’t, just stop using AI because it’s such, a high, productivity enhancer.
And so AI and the recent developments around cyber cost, the resilience landscape, the entire resilience landscape to change.
And I like to think of it as, three main factors. One is the attackers are now industrialized. It’s not just AI, but it’s a lot of it has to do with AI. If one as an attacker, you would have to put on your hoodie, go to a dark place and start reading, lines of code, or start, fuzzing, through code to find the vulnerabilities.
Now, you can use AI agents to do that for you at scale against a multitude of different platforms and applications and tools and find many vulnerabilities. It’s all it’s not all, but many of much of it is a matter of the tokens you have. You can buy vulnerabilities. You can buy credentials, from marketplaces. So the entire attack landscape is industrialized, and it’s much easier for attackers to carry out attacks that are more sophisticated, faster, and in wider scale against, organizations.
And we’ve all seen, the results of vulnerabilities like, PHP to shell, node to shell with the vulnerability and many, many other vulnerabilities, that came out lately as a result of AI. And the scary ones are the ones we haven’t heard about. Right? And obviously there are such, vulnerabilities the attackers did not publish.
The second thing is that the attack surface became complex and, expanded the multitude of identities. There’s a human user that’s using an agent. It’s using another agent that’s using another agent. The cascading, permissions of agent. Each agent can do different things. Each agent can, step out of the boundaries that were planned for that agent.
There are assets in different clouds in, on prem, in different places. All of this causes the attack surface to be complex and expanded, with a lot of, factors. That makes it make life easier for attackers. And makes it very hard to understand, where attacks originated from, because let’s say that an attacker attacked an endpoint of a user.
From there, they jumped to, an agent that runs on a cloud, that uses another agent that runs on a different cloud and is able to access data, that it’s in a different place and, and, and finding out what happened, is very difficult. And lastly, it’s a good thing, but, the recovery is now part of the CSO mandate, part of the security team mandate.
If, you asked a lot of CISOs five years ago if, backup and recovery was part of their responsibilities, they would argue that it’s, you know, it’s an i.t. Thing, but now recovery is, I would say becoming core of the CSO activity. So that is the change in the resilience landscape.
And so, for, for such, changes in the attack landscape and in the ownership landscape, we need to be very agile as an organization when we are defending organizations.
We need to know a lot of things. Right? We need to know where we have sensitive data. We need to know who is able to access that data, which users, which agents? Under which conditions? We need to know, what’s exposed, what’s risky, where we have risks in our data landscape, we need to know a lot of things right to in order to, protect our organizations.
Everything that gives us context to be able to defend it. And that’s the second part, which is adapt. We not only need controls in place to protect, obviously, but because this landscape is so fluid, so fast, so scalable, we need to be able to adapt quickly. We need these controls, to be able to apply changes to these controls based on the knowledge we just accumulated.
For example, if my account is compromised and we know that it’s compromised. We need to be able to, limit, restrict the sensitive data access of my entire team, perhaps. And we need to backup data, faster in asset that I may touch. We need to do a lot of different things based on what we have in the organization.
And finally, we always need to be able to restore, to return to our, minimal viable, company, to a trusted, state for organization, as an example, we know that there is sensitive data in a place where we did not have sensitive data.
Now, what we need to adapt is who can access this data under which conditions? Maybe we need to mask that sensitive data and so on and restore. We always want to be able to restore, especially this part of the critical, parts where we have sensitive data.
Same goes for compromised credentials. We know that, a credential is compromised. We need to be able to adapt for that. Like I mentioned before, and when we identify lateral movement within the organization, we need to adapt accordingly and always be able to restore.
And over to you, Yoav.
The cornerstone of AI is data. So AI is only as good as the data that you have. And you have to use a data to feed, your AI systems in order to really take advantage of it.
To address the challenges and the risk in, exposing data, or using data, in the AI era, we’ve built a portfolio, to address, to address those challenges.
The portfolio is comprised of two, offerings. The first product is called Vol Data Poster 360. That is our data security posture management solution. Security teams and compliance teams can use Data Posture 360 to discover and classify sensitive data across on prem and multi-cloud environment and understand where you have, risk in your data environment and, manage it and address it.
Commvault Data Access Manager, the product formerly known as Satori, helps DevOps teams help analytics teams, enforce real time security policies. When folks access data, help them and create just in time access workflows and, policies to ensure that sensitive data is protected. Let’s look at these two offerings, in a little bit more detail.
Data Posture 360 is the evolution of an existing product we already have called risk analysis.
From an architecture perspective, we’re combining risk analysis technology with Satori technology to deliver a unified data security posture management solution that, can, cover, classify data and help you manage data risk across, a very broad, data environment from SaaS applications and cloud workloads like this. There’s Azure SQL databases to your on prem file servers and databases like SQL or Oracle, and even your endpoints and VMs.
Data Posture of 360 helps you, for example, identify VMs where you may have files with credentials, just laying around. This could be a very, potent target for hackers to go and use these credentials to move laterally within your organization. It helps you identify where you have sensitive data. For example, it could be exposed.
You may have a spreadsheet with customer data in an S3 bucket or in a SharePoint site that is accessible globally. It helps you identify all of these cases and it helps you manage those risks remediated, close down those issues so you can ensure, first of all, you understand where you have sensitive data and what sensitive data you have, and to that, you’re able to monitor your data risk over time and minimize that.
The second product is Data Access Manager. That access manager is your enforcement layer or your access control layer over data for things like, you know, databases in the cloud like Azure SQL or Synapse, and also does Snowflake, Databricks, your on prem databases, MySQL, Mongo, all the all the different environments where you may have.
We support, including, your LMS Satori can intercept your, AI prompts and responses, identify sensitive data in prompts, and help you understand, how agents use, are used in your organization and secure that environment as well.
Satori helps enforce real time access controls, things like dynamic masking, row level security, with approval workflows. So only the right people have access to the data that they need. Time based access. So access is not provisioned and just left, open. Access is revoked after, you, you set period of time that you can define.
And, Satori supports both cloud and on prem workloads in a very easy deployment. That is, really, decoupled from your databases to ensure that, you know, if you have on prem applications or high performance, databases, you can enforce you can deploy that security layer in front of these databases without interfering your operations.
These products work together to both deliver the visibility layer and risk management layer so we can understand where you have sensitive data and manage that over time, but also helps you, enforce access controls and policies when people actually need to access data. A lot of solutions only focus on the visibility part. We provide you both visibility and control over your data, so you can activate the AI in your organization in a secure and safe way to you.
Jose.
All right. Sounds great. Thank you so much.
Now, what we’ve read, we’ve heard now from slides nine through 12 we covered the know and adapt. So know is Data Posture 360. Right. What data we have, where it lives, how sensitive it is.
You know who can reach it. And under what policy a restore is what happens when something goes wrong.
You know this slide is the art of the possible, not shipping architecture. Just so you know, right. We want to take the context of everything that we were just talking about today and trying to put it into perspective of what is possible for you to actually achieve.
Now, if you look here in the left column, you’ll see the different types of signals. Now, the signals don’t necessarily come from us. It could be an endpoint protection platform where if flags ransomware behavior or identity provider, you know that flags anomalous access to the sensitive data or our own threat scan which finds malware sitting in a backup.
Now the trigger comes for wherever the customer already watches, right? So no matter where it has, where you have it in general, we can actually put it into our console and be able to import it and look at it.
The point that I want to talk about here is that the trigger comes from wherever the customer already is watching this, okay, that nobody has to be logged into the com vault console for this to start. So this is where now if we look at the center, the actual previous slides actually pay off.
The agent does not have to go and ask what the data was or who could touch it. No, in a doubt already answered that it knows the effective host. You know, held regulated data. It knows which identities had access. The context is what lets it decide rather than just a layer. So the human approves in the agent executes.
It would be clear it is on here. Here’s because I want to make sure that everyone understands that having the human in a loop is also something is very important. Now, if you look here on the right, this is where the action is it’s not a ticket. This is where the four things can actually fire in whatever order the situation calls for.
Cleanroom is an isolated recovery environment spun up on demand for forensics without touching production. Cleanroom Recovery. You know what we have today is deception traps. You know, new decoys placed around the exposed data, new threat Y sensors mimics any asset and deploys in minutes. Now you can quarantine, access entitlements revoked or masked in real time. You know, through data access manager, clean recovery restored the last known good copy already scanned by thread scan and something that we don’t have to reinfect all over APIs.
Now, today this is a human right. Reading all of this alerts and driving these four consoles. So the art of the possible is that having an agent doing the orchestration and the human approving it. Now that is restore.
Thank you for joining us today. Please feel free to reach out to your Commvault representative to learn more about these technologies and more to help you secure your enterprise data activation for AI.