BeyondTrust Advances Unified Security for AI Agents and Workloads Amid Rapid Growth
In a significant development for cybersecurity, BeyondTrust has unveiled enhanced capabilities within its Pathfinder Platform, marking a pivotal moment in the effort to secure AI agents operating alongside human users and autonomous workloads across cloud and SaaS environments. This initiative emerges from new threat research conducted by BeyondTrust Phantom Labs, revealing that a substantial number of enterprises are unknowingly managing shadow AI agents with privileged access that evade detection and governance by security teams.
The Rise of AI Agents in Enterprise Environments
AI agents have transitioned from experimental tools to essential production workloads. They are now responsible for initiating API calls, utilizing credentials, deploying code, and accessing sensitive data. Many of these agents can be deployed within minutes on low-code platforms, often wielding privileges that rival or even surpass those of human administrators. In numerous enterprise settings, the number of machine and AI identities has already eclipsed that of human identities, significantly expanding the attack surface associated with privileged identities.
Despite this rapid growth, most organizations lack visibility into the number of AI agents operating within their environments, the extent of their access, and the potential consequences if one of these agents is compromised. This lack of insight highlights the urgent need for a unified approach to securing AI, human, and machine identities across diverse environments.
Marc Maiffret, Chief Technology Officer at BeyondTrust, emphasized the interconnected nature of these identities. He stated that “agentic AI is not an isolated problem. It’s a subset of the broader non-human identity landscape.” He further noted that organizations cannot secure agentic identities in isolation, as these agents are linked with human identities, machine accounts, secrets, and entitlements across various environments. The Pathfinder Platform aims to provide comprehensive visibility and security across this spectrum, underscoring the importance of managing privileged identities effectively.
Comprehensive Security Solutions for AI Workloads
The Pathfinder Platform distinguishes itself from other AI security solutions by addressing both aspects of the agentic AI challenge. It offers protection for AI coworkers and local agents on endpoints, as well as for AI workloads that operate autonomously across cloud infrastructures and SaaS platforms.
Key Capabilities of the Pathfinder Platform
-
Endpoint Privilege Enforcement for AI Coworkers: The BeyondTrust Endpoint Privilege Management (EPM) solution enforces least privilege and application control for AI clients operating on endpoints, such as Claude and ChatGPT. This ensures that AI tools can only execute actions permitted by policy, establishing a critical enforcement layer that is unique among agentic AI security vendors.
-
AI Agent Discovery and Risk Analysis: BeyondTrust’s Identity Security Insights® provides comprehensive discovery, classification, and posture auditing for agentic AI. This includes connector coverage for major enterprise AI platforms such as OpenAI, Google Vertex AI, Salesforce Agentforce, ServiceNow AI agents, and AWS Bedrock. Organizations benefit from automatic discovery, privilege path mapping, risk scoring, and shadow AI detection wherever agents are deployed.
-
Secrets Management for Autonomous Agents: BeyondTrust Password Safe® manages, rotates, and enforces just-in-time access for the secrets and API keys that power agentic workloads. When combined with Identity Security Insights, organizations gain end-to-end visibility from agent discovery to credential management, effectively eliminating the static credential exposure that can lead to persistent attack surfaces.
Addressing the Surge of Non-Human Identities
Data from BeyondTrust’s Identity Security Insights indicates a rapid acceleration in the growth of AI agents within enterprise environments. Over the past year, organizations utilizing the Pathfinder Platform have reported a staggering 466.7% increase in enterprise AI agents. Many of these agents are deployed through low-code platforms and automation frameworks that span endpoints, cloud infrastructure, and SaaS applications.
Maiffret pointed out that the critical question for security teams is not whether they have AI agents, but rather, “What can they access, what secrets are they using, and what happens if one gets compromised?” He explained that a single AI agent’s impact can extend across identity providers, cloud infrastructures, SaaS platforms, and on-premises directories. The Pathfinder Platform is designed to map these cross-domain privilege paths, providing visibility into how a compromised agent could escalate access throughout the entire environment.
Proactive Risk Assessment for AI Security
BeyondTrust has also launched its Identity Security Risk Assessment (ISRA), which offers organizations immediate visibility into AI agent risks as part of a broader identity security posture analysis. This assessment can connect across enterprise identity and AI agent infrastructures in under an hour, delivering findings within 24 hours. The results include a comprehensive inventory of all AI agents, shadow AI detection, cross-domain privilege path analysis, and risk scoring aligned with the MITRE ATT&CK framework, accompanied by prescriptive remediation guidance.
The integration of these capabilities reflects BeyondTrust’s commitment to addressing the evolving landscape of cybersecurity, particularly as organizations increasingly rely on AI technologies. As the prevalence of AI agents continues to rise, the need for robust security measures becomes more critical than ever.
For further insights into the implications of these developments, visit securitymea.com.
Keep reading for the latest cybersecurity developments, threat intelligence and breaking updates from across the Middle East.
