CyberArk launches privilege controls to secure AI agent identities

CyberArk has announced the general availability of its Secure AI Agents Solution, expanding its Identity Security Platform with new privilege controls specifically designed to secure AI agent identities.

The adoption of AI agents within organisations is increasing rapidly, as businesses look to automate tasks and enhance operational efficiency. These agents are emerging as a new class of privileged identity in the enterprise environment, bringing with them unique security risks. Threats include incorrect agent actions, unauthorised usage, and the possibility of agents being taken over by malicious actors, which are of particular concern when these agents are granted elevated privileges.

To address these challenges, the CyberArk Secure AI Agents Solution applies targeted privilege controls, ensuring AI agents receive only the access needed for specific tasks and only for the necessary period. This approach is intended to limit risk and help prevent unauthorised access, thereby supporting the safe scaling of AI-driven processes across enterprises.

"As organisations embrace AI agents, both builders and defenders must understand how identity-centric risks evolve when agents require elevated privileges. Without strong discovery, robust privilege controls, and comprehensive lifecycle management, organisations risk losing visibility and opening the door to catastrophic agentic attacks. CyberArk uniquely secures the full spectrum of identities - humans, machines and AI agents - by applying the right level of privilege controls, enabling innovation while maintaining security and compliance," said Matt Cohen, CEO, CyberArk.

Research cited by CyberArk indicates a significant rise in AI agent deployment. According to the company's recent CISO survey, AI agent adoption is expected to reach 76% of organisations within three years, but fewer than 10% currently have appropriate security and privilege controls implemented.

The report, titled "Securing Agentic AI: Identity as the Emerging Foundation for Defence," notes that nearly 40% of financial institutions and software companies have AI agents in production already. However, less than one in ten organisations have established agentic security controls such as risk registries and dynamic authorisation at scale. Two-thirds of Chief Information Security Officers (CISOs) in the financial services and software sectors consider agentic AI among their top three cybersecurity risks, with over one-third ranking it as the highest concern. The majority expect security around AI agents to drive additional cybersecurity spending in the near future.

AI agents operate autonomously, often necessitating privileged permissions to carry out their tasks. CyberArk warned that such privileges, if not properly monitored and restricted, could be misused or exploited, potentially resulting in severe business or regulatory issues.

Privilege controls

The CyberArk Identity Security Platform is designed to extend comprehensive privilege management across all identity types: human users, machines, and, now, autonomous AI agents. The Secure AI Agents Solution adds these protections to AI agents, using just-in-time access principles, least privilege, and ongoing session monitoring to manage potential risks throughout the agent's lifecycle.

This alignment of privilege controls aims to ensure consistent governance, security, and monitoring across every identity within the environment, fostering organisational innovation without compromising compliance obligations or security standards.

Key new features of the Secure AI Agents Solution include comprehensive AI agent discovery across SaaS, cloud, and developer environments, providing detailed profiles that list ownership, roles, and access privileges. The solution also enforces strict authentication requirements and least-privilege access principles, with zero standing privileges and full auditing of agent activity.

Continuous real-time monitoring is included to detect threats and unauthorised actions, triggering automated alerts and enabling quick responses. Lifecycle management tools govern AI agents from their creation through retirement, helping organisations keep pace with emerging regulations and support audit readiness.

CyberArk said its approach addresses the need for a privilege-focused methodology when integrating AI agents into enterprise environments. By securing the identities of AI agents alongside humans and machines, the company aims to reduce operational and security risks, while enabling broader adoption of AI-driven automation within regulated industries.