Executive summary:
Okta for AI Agents - Core has received DoW IL5 Provisional Authorization. This extends identity-first governance to Controlled Unclassified Information, allowing DoW program offices to onboard, protect, and govern agentic AI inside secure mission boundaries.
What is Okta for AI Agents - Core?
Okta for AI Agents - Core is an identity governance solution that has achieved a signed Impact Level 5 (IL5) Provisional Authorization. This milestone extends identity governance capabilities beyond FedRAMP High environments into the Department of War (DoW)’s most stringent tier short of classified systems—where Controlled Unclassified Information (CUI) lives.
Why does IL5 governance matter for agentic AI?
IL5 environments run under stricter authentication and audit requirements. Critically, for a market moving fast on agentic AI, they also have a much lower tolerance for the kind of orphaned, over-permissioned, or unauditable identities that AI agents are prone to creating.
That distinction matters because the agents showing up in DoW environments right now—code assistants wired into mission-adjacent repositories, procurement copilots touching CUI, autonomous agents chaining calls across Model Context Protocol (MCP) servers and mission systems—don't pause before crossing the IL4/IL5 line to check their own governance posture. If your identity fabric can’t govern agentic activity in IL5, scaling those agents at mission speed becomes near impossible.
What risks does ungoverned AI create in mission environments?
Running AI agents on any regulated data outside of a FedRAMP-authorized environment creates more than just a compliance gap—it creates a significant security threat with concrete risks. For IL5-level data, those risks don’t disappear—they compound:
- Compliance violations: At IL5, violations include an agent touching CUI data, possibly outside its authorized boundary.
- Compounding breach risk: A compromised credential compounds breach risk when it belongs to an agent operating in a mission-critical environment with machine-speed lateral movement across connected systems.
- Failed audits: The lack of agentic chain-of-custody evidence directly impacts your ability to field a system supporting the warfighter.
- Stalled AI adoption: Accreditation bottlenecks prevent program offices from deploying AI operational capabilities.
The Okta blueprint for IL5 agent governance
Okta for AI Agents - Core’s three-part blueprint—Onboard, Protect, Govern—enables agencies requiring IL5 to answer three key questions when managing their agentic presence: Where are my agents? What can they connect to? What can they do?
| Governance pillar | Core focus question | IL5 implementation details |
|---|---|---|
| Onboard | Where are my agents? | Centralizes agent discovery in an IL5-authorized Universal Directory inventory for all agents touching CUI or mission systems. |
| Protect | What can they connect to? | Replaces static keys and standing access with short-lived, scoped credentials for authorization servers, brokered consent for applications, and governed MCP server connections—providing the same Zero Trust posture DoW programs already require of human identities. |
| Govern | What can they do? | User access requests, certifications, agent deactivation (“kill switch”), and security information and event management (SIEM) audit log streaming generate evidence to support DoW accreditation and continuous monitoring requirements. |
For DoW program offices, this means the identity fabric you've already accredited for humans now extends natively to your AI agents and non-person entities, without a separate integration or a new authorization boundary to defend.
Secure your AI agents at the mission edge
DoW components and contractors currently operating Okta for US Military can access Okta for AI Agents - Core inside their existing IL5 boundary today. Documentation is available for your Authorizing Official’s immediate review to extend existing accreditation packages to cover agent identity security.
If you've been evaluating agentic AI capabilities but have been waiting for the identity layer to catch up before bringing it inside the wire, your wait is over. Okta for AI Agents - Core removes the friction, letting you innovate securely.
Contact our federal and public sector team to discuss your specific accreditation boundary and mission requirements.
