AI agents are creating new security challenges for enterprises as employees use them to interact with databases, cloud services, and other systems containing sensitive information. Security teams must determine which tools are running, what those tools are connecting to, and whether their access to enterprise data should be permitted. Bloom Security and Cyera are bringing endpoint monitoring and data classification together to help organizations make and enforce those decisions in real time.
Making Data Classification Enforceable
The integration connects Bloom’s visibility into AI activity on employee endpoints with Cyera’s understanding of enterprise data. Bloom identifies the AI agents and tools running on devices, the users operating them, and the destinations they connect to. Cyera supplies information about the data involved, including its sensitivity level, type, record count, and regulatory classification.
The combination is intended to help security teams move from identifying potentially risky AI activity to deciding what that activity should be allowed to do.
Consider an enterprise in which employees use AI agents that connect to cloud datastores or business applications. Knowing which agent initiated a connection is useful, but security teams also need to understand whether the destination contains personally identifiable information, protected health information, payment card information, credentials, or business records.
Bloom and Cyera bring these perspectives together. The resulting context can help teams prioritize activity according to the sensitivity of the data involved and apply policies that reflect the organization’s requirements.
Blocking Access While AI Agents Run
A central feature of the integration is its ability to extend Cyera’s data classification into runtime enforcement at the endpoint.
Organizations can establish policies that block AI agents from accessing datastores classified as Sensitive and above. Bloom then enforces those policies across endpoints as agents run.
This allows security teams to use data sensitivity as a basis for access decisions rather than relying only on the identity of an application or the presence of a connection. The companies describe the integration as enabling real-time decisions to block, pause, investigate, or remediate AI activity.
The approach also gives security teams a way to apply consistent controls across employee devices. Instead of treating data classification and endpoint security as separate processes, the integration connects information about sensitive data to controls operating where AI activity occurs.
The intended outcome is to stop risky AI activity before sensitive information leaves the organization.
Why Traditional Endpoint Visibility Is Not Enough
AI tools introduce a more complicated picture of endpoint activity. Employees may run dozens of AI agents, MCPs, and other tools, many of which connect to enterprise services and data repositories.
Monitoring installations and connections can show security teams what is happening on a device. However, understanding the potential impact requires additional information about the data those connections can reach.
Cyera provides that data context by identifying sensitive datastores and classifying the information they contain. Bloom supplies the endpoint perspective, identifying agents, users, devices, and connection targets.
Combining the two allows security teams to understand AI activity in relation to the sensitivity of the information involved. It also gives them a basis for investigating incidents with more context about potential data exposure.
Connecting Governance With Endpoint Controls
The integration reflects an effort to connect data governance decisions with the technical controls needed to enforce them.
Cyera’s platform helps enterprises understand what data they have and how it is classified. Bloom focuses on controlling AI tools and their associated software ecosystems on endpoints, using asset intelligence and information about users, running software, and accessible data.
Together, these capabilities allow data classification to inform endpoint policies. The companies say security teams can prioritize AI activity based on sensitivity, enforce policies consistently, and better understand data impact when investigating events.
“Data security has always been close to my heart, having spent a large part of my career building in this space,” said Itay Keren, CEO and co-founder of Bloom Security. “The challenge now is that AI agents are running on endpoints with direct access to sensitive data, and traditional endpoint security simply can’t see which agents are touching which data. Bloom monitors endpoint AI activity. Cyera knows what data is sensitive. Together, we’ve brought together endpoint visibility with data context: real-time enforcement grounded in data sensitivity.”
Moving From Policy to Protection
For enterprises adopting AI agents, a data classification policy is only one part of the security equation. Organizations also need to connect that policy to the systems where AI activity takes place.
Bloom Security and Cyera are addressing that requirement by linking data sensitivity with real-time endpoint enforcement. Their integration gives security teams a way to translate classifications into access restrictions, bringing data context into decisions about what AI agents can reach and helping organizations manage the risks associated with AI-driven workflows.
