IBM extends HashiCorp Vault to AI agents with Agent Identity and real-time revocation
IBM announced Agent Identity, Vault 2.2 and a kill switch in watsonx Orchestrate on 27 October 2026 to control what enterprise AI agents can access.
By Emre A.Published
Tech AI LatestIBM announced on 27 October 2026, at TechXchange in Atlanta, a set of controls for enterprise AI agents. HashiCorp Agent Identity records who owns each agent and what authority it has, while HashiCorp Vault issues short-lived credentials and can revoke a session in real time when an agent acts outside its permissions. watsonx Orchestrate adds discovery of agents built on Amazon Bedrock, Microsoft Foundry and Google Gemini, isolated sandboxes and a kill switch.
Key facts
- Announced: 27 October 2026 at IBM TechXchange 2026 in Atlanta
- Agent Identity: an authoritative record of each agent's owner, lifecycle status and granted authority, used with HashiCorp Vault
- Vault 2.2: separate secured accounts for agents, automatic credential renewal and post-quantum encryption
- watsonx Orchestrate: Discover and Contain in tech preview, with sandboxes and a kill switch for any agent
- NVIDIA: Agent Identity and Vault integrate with the NVIDIA OpenShell secure runtime
- Not stated: pricing and general availability dates for most features
What IBM announced
IBM used its TechXchange 2026 conference in Atlanta to introduce identity and control features for AI agents on 27 October 2026. The centrepiece is HashiCorp Agent Identity, which keeps an authoritative record of each agent: who owns it, its lifecycle status and the authority it has been granted. It shares that context with HashiCorp Vault, which IBM already uses to manage secrets and credentials for applications and services.
Vault now issues short-lived credentials that limit each agent to what its task needs. If an agent tries to retrieve data or take actions outside its approved permissions, IBM says access can be denied and the session revoked in real time. Vault 2.2 gives agents their own secured accounts, renews their credentials automatically and adds post-quantum encryption. A Secure Hub, in public preview, gives a central view across Vault environments.
Discovery, containment and NVIDIA
IBM watsonx Orchestrate gains three capabilities. Discover, in tech preview, connects to Amazon Bedrock, Microsoft Foundry and Google Gemini to find agents built in third-party environments, plus models and MCP servers. Contain, also in tech preview, runs agents in isolated sandboxes so one rogue agent cannot interfere with others, and adds a kill switch to shut down any agent. Evaluate tests agents, models and MCP servers before deployment and tracks each agent's model and token costs in production.
IBM also said Agent Identity and Vault integrate with NVIDIA OpenShell, a secure runtime for agents. NVIDIA vice president John Fanelli said the combination pairs a policy-governed runtime with verified identities and tightly scoped access to business systems.
Why it matters
Agents that call tools with long-lived, over-privileged keys are one of the clearest new attack paths in enterprise AI, because a hijacked or misbehaving agent can use whatever its credentials allow. IBM cites a Gartner forecast that the average Fortune 500 company will run more than 150,000 agents by 2028, and a finding that only 13% of organizations think they have the right agent governance. Treating each agent as a non-human identity with expiring credentials applies a known security pattern to a new kind of user. Similar agent runtime controls from other vendors are covered in Cybersecurity, including CrowdStrike Falcon Guardian.
What to watch next
Discover and Contain are tech previews, and IBM does not give pricing or general availability dates for Agent Identity in the release. Customers speaking at TechXchange include Albert Einstein College of Medicine, which IBM says is applying IBM Verify and Vault to agent identity and just-in-time access.
What changed
| Before | HashiCorp Vault managed secrets for applications and services, and agents typically reused service credentials |
|---|---|
| After | Agents get their own governed identity, short-lived Vault credentials and real-time revocation, with cross-platform discovery and a kill switch in watsonx Orchestrate |
| Who it affects | Security and platform teams running AI agents across Bedrock, Foundry, Gemini and in-house systems |
Questions people ask
- What is HashiCorp Agent Identity?
- IBM describes it as the authoritative record for each AI agent, covering its owner, lifecycle status and granted authority. It feeds Vault's credential controls to enforce least-privilege access.
- Can IBM's tools stop an agent mid-task?
- IBM says Vault can deny access and revoke a session in real time when an agent acts outside its permissions, and watsonx Orchestrate adds a kill switch for any agent.
- Which agent platforms can watsonx Orchestrate discover?
- Amazon Bedrock, Microsoft Foundry and Google Gemini, according to IBM. The Discover feature is in tech preview.
Sources
- IBM Introduces Identity and Control for AI Agents, Securing Frontier Intelligence for the Enterprise (IBM Newsroom, 27 Oct 2026)
Reference pages
Spotted an error? See the corrections policy and tell the editor. Facts in this story come from the sources listed above; vendor claims are attributed.
Comments
Loading comments.
