Skip to content

Add guidance for controlling AI agent identity and permissions #1422

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
azarboon wants to merge 2 commits into
base: main
Choose a base branch
from
Open

Add guidance for controlling AI agent identity and permissions #1422

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
dc9380e
Enhance AI agent security guidelines
azarboon Jan 18, 2026
80969f8
Update secure.md
azarboon Jan 18, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 2 additions & 0 deletions docs/scenarios/ai/secure.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -42,6 +42,8 @@ Comprehensive security measures protect your AI investments and maintain stakeho
- [Azure PaaS security](./platform/security.md)
- [Azure IaaS security](./infrastructure/security.md)

4. **Control AI agent permissions and autonomy.** AI agents with autonomous capabilities can perform actions that exceed intended authorization boundaries, introducing security and compliance risks. To mitigate excessive agency, as described in [OWASP LLM06](https://owasp.org/www-project-top-10-for-large-language-model-applications/), represent AI agents as first-class identities using [Microsoft Entra Agent ID](https://learn.microsoft.com/entra/agent-id/identity-professional/microsoft-entra-agent-identities-for-ai-agents) where supported. Use [Azure role-based access control (RBAC)](https://learn.microsoft.com/azure/role-based-access-control/overview) to scope agent permissions to the minimum set of resources required for their tasks, implement [human-in-the-loop (HITL)](https://learn.microsoft.com/agent-framework/tutorials/agents/function-tools-approvals) checkpoints for high-impact operations such as financial transactions or data modifications, and, where available, apply [Conditional Access](https://learn.microsoft.com/entra/identity/conditional-access/overview) policies to evaluate agent-initiated activity in alignment with [Zero Trust principles](https://learn.microsoft.com/security/zero-trust/zero-trust-overview).

### Secure AI data

AI workloads rely on data and artifacts that require robust protection to prevent unauthorized access, data leaks, and compliance violations. You must implement comprehensive data security measures to protect AI data and artifacts. Here's how:
Expand Down