ServiceNow, which is billing itself as “the fastest-growing major enterprise cybersecurity company,” is bringing identity, exposure management and incident response into a unified security offering as enterprises give AI agents greater access to business systems and data.
The vendor is expanding its cybersecurity portfolio with six groups of security capabilities designed to help enterprises manage the risks created by increasingly autonomous AI systems.
The announcement comes as enterprises move AI agents from information retrieval and assistance toward tasks that involve accessing applications, modifying data and triggering business processes. This is creating a security challenge that traditional controls were not designed to address.
An AI agent can have its own credentials, access multiple applications, consume untrusted information and make decisions about what action to take. If compromised or poorly governed, that agent could provide an attacker with a route into systems that previously required a human user to access.
Cybersecurity experts are warning that enterprises need stronger approaches to identifying, authorizing and blocking AI agents as they gain access to diverse datasets, tools and applications.
ServiceNow is positioning its new Autonomous Security portfolio around the idea that security teams need to govern assets, identities and AI agents through the same operational workflows.
AI Agents Create a New Identity Problem
One of the significant elements of ServiceNow’s announcement is its focus on non-human identities.
Traditional identity and access management has largely been designed around employees, contractors and other human users. AI agents add another category of identity, with potentially different lifecycles, permissions and behavioral patterns.
An agent may need access to a CRM system to retrieve customer information, for example, while another could have permission to update records or initiate refunds. If those permissions are broader than necessary, or persist after the agent’s task has finished, they create additional exposure.
The problem becomes more complicated when agents interact with other agents and applications, creating chains of automated activity that can be difficult for security teams to reconstruct.
ServiceNow’s AI Agent Access Security is intended to provide access controls for AI agents across different platforms and model providers. Its Non-Human Identity Remediation capability is designed to automate actions including key rotation, deprovisioning and permission revocation across IT, OT, IoT and medical environments.
This direction is consistent with a broader industry focus on giving agents explicit identities and permissions. Microsoft, for example, has argued that securing an agentic workforce requires treating agents as identities within an organization’s security architecture.
ServiceNow Expands Exposure Management
ServiceNow said its Agentic Exposure Management capability will consolidate vulnerability findings from multiple sources and enrich them with threat intelligence and business context.
It is also introducing a Vulnerability Resolution AI Specialist that can orchestrate vulnerability triage and remediation, including executing low-risk patches.
That distinction is important as enterprises generate more software and infrastructure through AI-assisted development. The volume of code, dependencies and potential misconfigurations is increasing, while security teams are under pressure to prioritize vulnerabilities based on real business impact rather than treating every issue as equal.
ServiceNow is extending its Application Security capabilities to cover AI-generated code and model dependencies, alongside Dynamic Application Security Testing and External Attack Surface Management.
The objective is to connect vulnerabilities across application, infrastructure and external attack surfaces rather than treating each as a separate security problem.
Autonomous Response Moves Into the SOC
ServiceNow is also applying AI agents to security operations. Its Agentic Incident Response capability includes a Tier 2 SOC AI Specialist designed to investigate incidents and execute multi-stage response plans.
The system can perform activities such as enrichment, correlation, containment and blocking, with higher-risk decisions escalated to human analysts.
Security operations are shifting toward AI-assisted investigation and response as the volume of alerts and security telemetry increases. Security teams are looking for ways to automate repetitive investigation work, reduce analyst fatigue and respond faster to active threats.
The challenge is that giving an AI system the authority to contain an incident introduces another layer of risk. The organization needs to know what the agent can change, why it made a decision, what data informed that decision and how the action can be reversed.
Mary Ann Miller, Fraud and Cybercrime Executive Advisor and VP of Client Experience at Prove, noted in a CX Today panel discussion that auditability becomes essential when AI agents are operating autonomously or semi-autonomously.
“Are you able to connect the dots of the decisions that it’s making, the actions, and do you need to audit that?”
That question goes to the heart of autonomous security operations. AI-driven investigation may reduce response times, but security teams still need traceability, approval thresholds and mechanisms to intervene when the system’s behavior changes.
ServiceNow says its approach is intended to provide that governance and auditability through its existing workflow and orchestration infrastructure.
Security Extends Into Operational Environments
The company’s portfolio also extends into cyber-physical environments.
ServiceNow’s Agentic AI for Cyber Physical Security is designed to provide visibility across OT and medical networks without deploying traditional agents. It can establish behavioral baselines, monitor compliance and model potential attack paths.
Drawing on technology from its acquisitions of Armis and Veza, the combination gives ServiceNow visibility into two important aspects of enterprise security: what is connected and who or what can access it.
That matters as AI agents increasingly operate across conventional IT systems and physical or operational environments. In sectors such as healthcare, manufacturing, utilities and logistics, the consequences of security failure can extend beyond data exposure into operational disruption.
The ability to map connected assets, monitor behavior and understand access relationships is therefore becoming more important as AI-driven workflows move closer to critical systems.
Compliance Moves Toward Continuous Monitoring
ServiceNow is also targeting the compliance implications of autonomous systems.
Its Agentic AI for Continuous Control Monitoring is designed to continuously evaluate areas including segregation of duties, access rights and configuration states across ServiceNow and external systems. The company said the system can identify control violations as they occur rather than waiting for periodic audits.
A separate Cryptographic Asset Compliance capability is intended to help organizations identify legacy cryptographic technologies and plan migration toward quantum-resistant standards.
The continuous monitoring direction reflects a broader shift in enterprise risk management. Periodic audits are poorly suited to environments where AI agents can make decisions, access systems and trigger workflows in real time. If the underlying access state changes, or if an agent begins behaving unexpectedly, enterprises need faster detection and response.
The Security Question Changes as Agents Gain Authority
ServiceNow’s release reflects a wider change in enterprise AI security. The central issue is increasingly becoming an authorization and execution problem.
Conventional security principles remain relevant, but they need to be adapted to account for the characteristics of AI agents. For enterprises, AI governance increasingly needs to answer practical security questions around agents’ identities and access, alongside questions about model performance and responsible use.
That includes:
- What identity does the agent use?
- What systems can it access?
- What actions can it perform?
- Can permissions be revoked automatically?
- How are decisions and actions logged?
- What happens when behavior deviates from the expected pattern?
- When does the workflow return to a human analyst?
Geoffrey Mattson, CEO of SecureAuth, summarized the practical control model for agentic systems:
“Scope down what they’re allowed to do, monitor it in real time, and look for drifts… and then always be able to get back to a human.”
That framing aligns with the direction of ServiceNow’s Autonomous Security portfolio: access controls, exposure management, continuous monitoring, incident response and human escalation brought into a common operational layer.
Whether that approach can reduce the complexity created by the proliferation of AI agents will depend on how broadly enterprises can connect their existing security and operational systems to it.
ServiceNow said most of the capabilities are available now. The Tier 2 SOC AI Specialist, Vulnerability Resolution AI Specialist, Agentic AI for Continuous Control Monitoring and Cryptography Asset Compliance are scheduled for release in December 2026.
