Keep enforcement close, human authority clear, and every governed action on the record.
RSA Agent ID
Your agents can act for you. Make sure they answer to you. Put RSA Agent ID to work across the enterprise and make high-assurance identity the foundation of your agentic security.
Know every agent. Control every action.
The agentic identity security platform for highly-regulated industries.
Sovereign control
Run Agent ID in the cloud, hybrid, or on-premises, chosen per gateway.
Know every agent
Discover sovereign and sanctioned agent
Control every action
Enforce your policy on each call at the AI/MCP Gateway
Autonomy, meet accountability.
The age of improvised agentic security is over. RSA Agent ID makes every agent known, every permission enforceable and every critical action answerable to a human. So you can hold every agent to the highest level of assurance.
Agentic security will be won or lost on identity.
Agent autonomy has outrun identity security. They’re holding credentials. Armed with permissions. Making decisions at machine speed. Most organizations have lost count, lost sight, and lost the chain of command. The identity crisis is already unfolding.
- 1
- 10001
- 20001
- 30001
- 40001
- 50001
- 60001
- 70001
- 80001
- 90001
- 100001
- 110001
- 120001
- 130001
- 140001
- 150000
AI agents the average Fortune 500 enterprise is projected to run by 2028, up from fewer than 15 in 2025.
- 1%
- 6%
- 11%
- 16%
- 21%
- 26%
- 31%
- 36%
- 41%
- 46%
- 51%
- 56%
- 61%
- 66%
- 71%
- 74%
Of IT leaders say AI agents create a new attack vector into their organization.
- 1%
- 2%
- 3%
- 4%
- 5%
- 6%
- 7%
- 8%
- 9%
- 10%
- 11%
- 12%
- 13%
Of organizations believe they have the right AI agent governance in place.
Find them. Secure them. Govern them.
Built for complexity. Proven at scale. Designed for organizations where the generalists aren’t good enough.
Discover
What agents are operating in your environment?
Find agents and MCP servers, sanctioned and shadow, across supported identity, cloud, endpoint and Gateway sources. Bring them into one registry with:
- A named owner from the day each agent is found
- A classification, risk tier and lifecycle state
- Correlation to the identity provider you already run
Secure
Who is accountable for each?
Apply policy to each call routed through the AI/MCP Gateway:
- Control access at the tool and argument level
- Make sure delegated authority can only narrow
- Require a named, authenticated human to approve actions you mark high-risk
- Revoke access when an agent is decommissioned
Govern
Can anyone stop them?
Minimize privilege, certify access, and deliver evidence for ten industry frameworks.
- Continuous certification
- Risk-based access reviews
- Lifecycle automation
Every Identity. Secured.
RSA ID Plus
invisible
invisible
Secure access across cloud, hybrid and on-premises environments with identity and access management solutions built for high-assurance.
RSA Governance & Lifecycle
Bring visibility, accountability and lifecycle control to access across your organization.
RSA SecurID
invisible
invisible
Protect critical access with phishing-resistant authentication trusted in the world’s most demanding environments.
RSA Risk AI
invisible
invisible
Use behavioral intelligence to identify risk and inform access decisions as conditions change.
Make autonomy answer to you.
Sovereign Control
Sovereignty stays with you
Sovereignty stays with you
Put the RSA Agent ID Gateway where your requirements demand. Host it in your environment and keep its decisions, keys and evidence on your side. Choose the region for tenant data. Go fully air-gapped and self-managed in 2027.
Provable Human Authority
Humans hold the line
Humans hold the line
Every agent starts with an owner. Every high-risk action ends with a verified human decision. RSA Agent ID preserves the chain of command from authorization to action.
Regulatory Evidence
Regulators, ask away
Regulators, ask away
Every governed action builds the record for you. RSA Agent ID captures the agent, the rule, the human, the tool, and the delegation chain. Then it maps that evidence to ten major frameworks and sends it to your SIEM. So you can meet every examiner with evidence, not explanations.
Go deeper on agentic identity.
To secure agents, secure identity.
Download the white paper to learn the identity controls regulated industries need to stay secure in the agentic era.
RSA Agent ID: The identity standard, extended to agents
Get the complete Agent ID announcement.
Every agent must answer to someone
RSA CEO Greg Nelson on why RSA is bringing a new standard to agentic security.
Build on your identity provider. Not around it.
Keep the provider. Make agent authority enforceable. RSA Agent ID works with RSA ID Plus, Microsoft Entra, AWS IAM, Microsoft Defender, and more, and uses the credentials they already use. Your provider establishes authority. Agent ID makes sure agents stay within it.
Frequently asked questions
What is RSA Agent ID?
RSA Agent ID is an agentic AI security platform for government, financial services, and other highly-regulated organizations. It discovers AI agents and MCP servers, assigns each one an accountable human owner, enforces policy on every call through an AI/MCP Gateway you can deploy inside your own boundary, and records audit evidence of every agent action.
When does RSA Agent ID ship?
Discover and Secure ship November 16, 2026. Govern ships early 2027.
How does RSA Agent ID find shadow agents?
RSA Agent ID Discover correlates multiple discovery signals to find sanctioned and unsanctioned agents and MCP servers. It registers each one with a named owner, risk tier, and lifecycle state tied to your identity provider. IBM’s 2026 Cost of a Data Breach Report puts the average shadow AI incident at $5.39 million.
What governance is available before Govern ships?
Discover and Secure establish the governance foundation at launch: classification, risk tiering, mandatory ownership, lifecycle states and evidence mapping. Continuous certification, risk-based access reviews and lifecycle automation arrive with Govern in early 2027.
Does RSA Agent ID replace my identity provider?
No. RSA Agent ID integrates with RSA ID Plus, Microsoft Entra ID, Okta, AWS IAM, CrowdStrike, Microsoft Defender, and more today, with additional agent platform integrations planned.
What is an AI/MCP Gateway?
The Model Context Protocol (MCP) is an open standard that lets AI agents call external tools and data sources. An MCP gateway sits between agents and those tools and checks each call against policy before it runs. RSA Agent ID Secure runs this gateway RSA-hosted or in your private cloud, on-premises, or air-gapped environment.
Where can I run the Gateway?
Gateways can be RSA-hosted or customer-hosted in cloud, hybrid or on-premises environments. A fully air-gapped, self-managed version is coming in 2027.
Where does my data live?
Tenant data stays in the region you select: United States or EMEA. Enforcement evidence is generated where the Gateway runs.
Which actions require human approval?
You determine which actions are high-risk. Those actions wait for approval from a named person, verified against your identity provider and approved with a phishing-resistant credential.
How does RSA Agent ID support my compliance program?
RSA Agent ID maps evidence to ten frameworks:
- NIST AI RMF 1.0
- ISO/IEC 42001
- Treasury Financial Services AI RMF
- NYDFS Part 500
- DORA
- NIST SP 800-53 Rev. 5
- OMB M-25-21 and M-25-22
- FedRAMP
- EU AI Act
- OWASP Agentic AI Top 10
What happens to an agent the moment Discover finds it?
It’s added to one registry with a named owner, a classification, a risk tier and a lifecycle state—assigned on discovery, not after.
Does Discover treat sanctioned and shadow agents differently?
No. Both land in the same registry, correlated to the identity provider you already run.
Who owns an agent once it’s discovered?
Every agent gets a named owner from day one. None enter the registry unowned.
What does Secure control beyond who can approve an action?
Access at the tool and argument level—policy governs not just whether a call goes through, but what parameters it carries.
Can an agent’s permissions grow over time?
No. Delegated authority can only narrow as it passes from one call to the next, never broaden.
What happens to an agent’s access when it’s decommissioned?
Access is revoked the moment the agent is decommissioned. Nothing lingers.
Does Govern apply only to AI agents?
No. It extends RSA Governance & Lifecycle to employees and agents on one platform—same certifications, same reviews.
What does continuous certification mean under Govern?
Access is certified on an ongoing basis, not on a fixed schedule, so entitlements track actual risk in real time.
How does lifecycle automation build on the lifecycle states Discover assigns?
Discover assigns and tracks the state. Govern automates what happens next—certifying, revoking, or escalating access as that state changes, without manual work.
Skip the promises. See the proof.
Get a firsthand look at RSA Agent ID and see what happens when agents answer to a higher standard.
Included