BeyondTrust unveils AI security tools to govern agentic AI risks

BeyondTrust has introduced new artificial intelligence (AI) security controls in its Identity Security Insights platform that aim to provide detailed oversight and governance for AI agents alongside traditional human and service identities.

The company has identified agentic AI – autonomous and persistently operating AI systems capable of making decisions and executing actions – as a significant development in enterprise technology, with both substantial efficiency opportunities and potential cybersecurity risks. BeyondTrust's latest capabilities are designed to address this emerging challenge by extending identity security controls to encompass these AI entities.

Agentic AI concerns

Agentic AI systems differ from earlier algorithms by being able to take independent actions such as scheduling meetings, provisioning resources, or interacting with other digital identities, rather than just offering alerts or recommendations. This shift, according to BeyondTrust, heightens the risks facing enterprises if these agents are compromised or misused.

The company's announcement draws attention to scenarios in which artificial agents may act without the explicit consent or awareness of their owners, potentially leading to harmful mistakes or exploitation by threat actors. The risks are amplified by the capacity of AI agents to operate with significant privileges, sometimes rivaling those of human administrators.

Expanded security capabilities

BeyondTrust's update to Identity Security Insights delivers three core capabilities intended to secure the use and governance of agentic AI:

• AI Agent Insights: This feature allows customers to discover, classify, and risk-assess AI agents across cloud and SaaS platforms, including applications such as Salesforce Agentforce and ServiceNow. It also identifies unauthorised or 'shadow' AI usage and applies governance controls like Zero Standing Privilege and Just-In-Time access policies.
• Model Context Protocol (MCP) Orchestration: Using BeyondTrust's lightweight MCP Server, this creates a secure mechanism for brokering actions between AI agents and enterprise systems, ensuring workflows such as just-in-time API requests and credential management are executed with appropriate security controls.
• Omnipresent AI Decision-Support Layer: Embedded in the Pathfinder Platform, this assistant provides real-time insights and guidance on identity security, helping teams to quickly interpret data and take action without leaving existing workflows. This feature is powered by BeyondTrust Phantom Labs research.

Industry viewpoint

"The rise of AI agents is creating a new and urgent identity security challenge. Agentic AI is not an isolated problem; it's a subset of the broader non-human identity landscape. To secure it effectively, organizations must think holistically about their entire identity ecosystem, not just one type of identity in a silo. Often built on low-code and no-code platforms, AI agents can be deployed in minutes with privileges that rival human admins. BeyondTrust uniquely connects visibility with proactive control across all identities so customers can rein in this new frontier of risk and turn AI into a safe force multiplier that also meets compliance requirements."

said Marc Maiffret, Chief Technology Officer at BeyondTrust.

Addressing shadow AI

BeyondTrust noted that many organisations underestimate the extent to which agentic AI is already present within their networks. To assist in identifying these hidden risks, the company has expanded its Identity Security Risk Assessment service to provide immediate visibility into both registered and shadow AI agents as part of its wider identity security framework.

The company said that these new features build on previous platform advancements, such as the introduction of Secrets Insights for managing risks associated with secrets and non-human identities. Adding AI agent governance is described as an extension of BeyondTrust's mission to unify identity governance across the entire attack surface, encompassing human users, service accounts, secrets, and now, AI and machine identities.

Customer benefits outlined

For organisations seeking to embrace AI adoption while maintaining strong security and compliance, BeyondTrust asserts that its expanded platform permits faster AI integration with consistent visibility, governance, and enforcement across all identity types. This, in turn, is intended to reduce operational risk, improve team productivity, and simplify security management through a single solution.

The integration of these controls into BeyondTrust's Pathfinder Platform and Identity Security Insights is positioned as a step towards helping enterprises manage the increasing complexity of digital identities and ensure that rapidly deployed AI agents do not introduce unmanaged risk or violate compliance requirements.