Published: November 4, 2026 | News development: November 3, 2026
IBM has made Agent Identity generally available in watsonx Orchestrate, giving enterprise AI agents individual identities, access records and a one-step shutdown mechanism. The update addresses a growing challenge for organizations that are moving AI agents from demonstrations into workflows that interact with business applications and sensitive data.
What IBM Agent Identity does
IBM says each agent running in watsonx Orchestrate can have its own verifiable identity registered through an organization’s existing IBM Verify or Microsoft Entra environment. The capability covers agents built in watsonx Orchestrate as well as external agents imported into its runtime, including LangGraph agents.
Administrators can use the identity view to see which agents exist, who owns them and what systems they can access. IBM says the design allows enterprises to use identity-management processes they already operate, rather than maintaining a separate identity store just for AI agents.
Audit logs show which agent acted and on whose behalf
A key addition in the general-availability release is agent-specific audit logging. Each access event records details such as the connection and tool involved, the user for whom the agent acted, and identifiers that help security teams trace the activity. The records can also be passed to existing security tools for investigation.
This distinction is important in an enterprise environment: an investigation may need to establish both which automated agent performed an action and which human user it was acting for. IBM says the approach uses an OAuth token exchange with a token scoped to the particular user and agent.
A kill switch can stop an agent in a live environment
IBM has also added a kill switch that can undeploy an agent from a live environment and end its active sessions in one step. Before shutdown, watsonx Orchestrate displays collaborator agents that may be affected, helping administrators understand potential workflow disruption.
Why AI agent identity matters for businesses
As companies adopt enterprise AI automation, agents may access customer records, internal documents, support systems and operational software. Shared credentials and unclear ownership make it difficult to apply least-privilege access or investigate unexpected actions.
- Security teams: can review agent ownership, permissions and activity.
- IT administrators: can use existing identity providers and established provisioning processes.
- Compliance teams: gain more traceable records of automated actions and the users associated with them.
Availability and setup
IBM says Agent Identity is available now in watsonx Orchestrate. Administrators can enable it under Govern & operate by connecting an IBM Verify or Microsoft Entra tenant. Organizations should review the product documentation and test the configuration against their own identity, access and audit requirements before enabling agents in production.
Read more technology coverage on NewsTree24, including updates on AI software, cybersecurity and data analytics.
Source
IBM: Agent Identity in IBM watsonx Orchestrate is now generally available, November 3, 2026.
Featured image: illustrative enterprise security image by path digital via Unsplash.