IBM Unveils Agent Identity and Security Tools to Control AI Agents Across Business Systems

IBM Tackles AI Agent Management with New Identity Framework

IBM announced a comprehensive suite of capabilities designed to address the governance, security, and operational control of AI agents at TechXchange 2026 in Atlanta1

1

2

2

. The company introduced Agent Identity through HashiCorp Vault, establishing what it describes as an authoritative record for each AI agent including ownership details, lifecycle status, and granted authority. This marks IBM's strategic response to the growing challenge organizations face as AI agents proliferate across business systems without clear accountability structures.

The announcement centers on a fundamental problem: as companies deploy AI agents across their infrastructure, they need mechanisms to verify who owns each agent, control what it can access, and trace its actions. IBM's approach combines agent discovery, verified identities, access controls, and performance evaluation into what the company calls an AI operating model built on four essential systems: agents, data, automation, and hybrid infrastructure2

2

.

HashiCorp Vault 2.2 Extends Credential Controls to AI Agents

HashiCorp Vault, which traditionally manages secrets and credentials for applications and services, now extends these capabilities specifically to AI agents through version 2.21

1

. The system issues short-lived credentials that restrict each agent to only what it needs, enforcing least-privilege access principles. When paired with Agent Identity, Vault establishes the definitive record for each agent and shares relevant identity and authorization context with its credential controls.

This integration enables real-time revocation when an agent attempts unauthorized actions. Vault 2.2 provides agents with their own secured accounts, automatically renews their credentials, and incorporates post-quantum encryption to address quantum-era threats2

2

. The short-lived nature of these credentials limits both what an agent can access and the duration of those permissions, reducing the window of exposure if credentials are compromised.

Secure Hub, now in public preview, adds centralized visibility across Vault environments, making it easier for organizations to understand operational health, usage patterns, and security resources across their AI agent deployments1

1

.

Watsonx Orchestrate Brings Discovery and Containment Capabilities

IBM watsonx Orchestrate now functions as a centralized hub for AI agent management, working across the AI ecosystem to discover, secure, and evaluate agents regardless of where they were built2

2

. The platform connects to third-party environments including Amazon Bedrock, Microsoft Foundry, and Google Gemini to automatically find agents, eliminating the need for manual tracking and registration.

The tech preview features include isolated sandboxes that prevent rogue agents from interfering with other agents or disrupting operations. Teams gain access to a kill switch that can immediately shut down any agent showing problematic behavior2

2

. This containment capability addresses a critical concern as organizations scale their AI agent deployments: the risk that a malfunctioning or compromised agent could cascade failures across interconnected systems.

Orchestrate now includes testing tools that evaluate the quality, accuracy, and compliance of agents, models, and MCP servers before deployment, then continue monitoring them in production. Built-in cost tracking shows each agent's model and token usage, helping teams identify what drives spending and optimize accordingly2

2

. The platform integrates with Vault's credential controls and Agent Identity to establish each agent's role, govern its authority, and enforce least-privilege credential access.

Integration Intelligence Hub and Infrastructure Context

IBM Integration Intelligence Hub provides agents with knowledge of existing APIs, tools, and system connections through its Unified Context Engine1

1

. This allows agents to address issues like stuck messages or deploy APIs using infrastructure already in place, rather than requiring new integrations. The system includes skills, approval policies, and audit controls to govern these actions2

2

.

Agent Identity and HashiCorp Vault integrate with NVIDIA OpenShell secure runtime software, which forms the foundation of the NVIDIA OpenShell Architecture. This integration pairs a protected environment for agents to operate within alongside verified identity and short-lived credentials for every agent action, ensuring each operation can be traced, limited, and reviewed1

1

2

2

.

Observability and Infrastructure Management Enhancements

IBM Concert integrates observability, optimization, incident management, and security across applications, infrastructure, and networks1

1

. HashiCorp Terraform provides infrastructure context and governed workflows for implementing changes. New features now available include Terraform policy, Terraform Developer Portal, and Terraform remote MCP server, enabling developers and AI agents to access approved configurations and implement infrastructure changes through established controls2

2

.

IBM Z Atlas, currently in tech preview, maps a client's entire z/OS stack across infrastructure, middleware, and applications. The system uses this intelligence to address manual tasks like patch management, regulatory response, and sharing system knowledge across teams from systems programmers to application developers2

2

.

HashiCorp Vault 2.2, Agent Identity, watsonx Orchestrate evaluate capabilities, IBM Concert, and HashiCorp Terraform policy are generally available. Secure Hub, Terraform Developer Portal, Terraform remote MCP server, and IBM Z Atlas remain in public preview. IBM Integration Intelligence Hub and watsonx Orchestrate discover and contain capabilities are in tech preview1

1

2

2

.