Protecting AI & LLMs within Cloud Apps

Through its ITP and PCCE capabilities, the Delinea Platform detects and rates the risks levels of deployed LLMs and AI agents throughout your cloud infrastructure.

Visibility into AI Models Deployed in CSPs

The Delinea Platform gives visibility into all active AI models, including their usage, ownership, and deployments— enabling alignment with organizational compliance and legal requirements.

Delinea incorporates publicly available model reputation indicators, including:

  • Hallucination rates

  • Fairness or privacy ratings

  • Safety thresholds

This additional context supports better decision making and risk analysis in case the LLM has access to sensitive assets.

Visibility into AI agents and services managed in CSPs

AI agents are applications or services that leverage AI models (LLMs) to perform tasks, respond to user queries, or automate operations. These agents are often deployed and managed within the cloud itself or run independently on cloud assets like VMs or containers.

The Delinea Platform provides deep discovery and visibility into AI agents and services hosted and managed in Azure, including rich metadata such as:

  • Underlying model and publisher

  • Creation date and region

  • Agent instructions and descriptions

  • Associated tools (e.g., Azure Functions, API calls, files)

The Delinea Platform analyzes potential access from AI agents and alerts those with access to sensitive assets.

Visibility into AI Models Hosted on Cloud Assets

The Delinea Platform extends visibility beyond managed services to detect self-hosted or "under-the-radar" AI models, such as LLMs deployed on Azure VMs, containers, or Kubernetes clusters (AKS). This helps organizations discover AI deployments that may bypass centralized governance and identify assets that can potentially run AI without your organization’ awareness.

Access Control and Risk Mapping

The Platform analyzes AI agents’ configuration, including their tools, functions, and instruction sets, to:

  • Identify potential access to sensitive or production assets

  • Flag agents with access beyond their intended scope

  • Support governance and least privilege enforcement

AI Reports for Governance

  1. Navigate to Insights > Reporting

  2. Enter AI agents to generate a report listing all known AI agents across your Azure environment.

This centralized view supports audit, compliance, and policy review.

Security Checks and Risk Assessments

The Delinea Platform runs automated risk assessments on AI & LLMs to improve IAM hygiene, including:

  • LLMs operating with high temperature (risk of unpredictable output)

  • AI agents lacking audit logs

  • AI services exposed to the public internet

  • Agents created by external accounts