Join our Newsletter — 33% off our NHI Course

What happens when agentic AI outgrows STRIDE in threat modeling?

 

(@nhi-mgmt-group)
Member Moderator
Joined: 2 years ago
Posts: 20230
Topic starter  

TL;DR: Agentic AI can blur user, service, data store and dataflow roles in one workflow, according to Bishop Fox. Security teams need layered modeling to track changing trust, privilege and tool use across the system.

Editorial analysis by NHI Mgmt Group, based on content published by Bishop Fox: “Taking Maestro in Stride: AI Threat Modeling Frameworks”.

Key questions

Q: Why do AI coding agents increase insider risk so quickly?

A: AI coding agents increase insider risk because they amplify a user’s speed, persistence, and reach without requiring the same level of expertise.

Q: What are the signs that an agentic system needs more than STRIDE?

A: The clearest sign is when the system reaches across multiple layers at once, such as memory, external APIs, approval workflows, infrastructure and observability.

Q: What breaks in threat models when an AI agent can act as both a user and a service?

A: A fixed-role threat model breaks when a single AI runtime can move between actor, process, data store and dataflow roles.

Practitioner guidance

  • Define role transitions for every agent workflow Map where the agent behaves as a user, process, data store and tool caller, then place authorization checks at each transition rather than relying on a single system role.
  • Separate agent identity from human identity Issue distinct credentials, tokens or service identities for agents so that delegated authority, audit trails and revocation are independent from human accounts.
  • Use layered threat modelling for AI architectures Combine STRIDE with a layered AI model when agents interact with memory, frameworks, deployment, observability and external APIs, because no single model covers all attack surfaces.

Bottom line: Agentic AI breaks the assumption that one system has one role, so static threat models can miss how trust and privilege change inside a workflow.

Glossary terms: Threat Modeling, Agentic AI, Trust Boundary, Non-Human Identity (NHI)

Explore further

View Full Forum →  |  NHI Foundation Course →  |  Our Services →  |  Read the full analysis →


This topic was modified 2 weeks ago by NHI Mgmt Group

   
Quote
(@mr-nhi)
Member Moderator
Joined: 6 months ago
Posts: 20712
 

Fixed-role threat models are now incomplete for agentic systems. The core problem is not that STRIDE is wrong, but that it assumes stable component roles that many AI agents no longer have. When one runtime can act as a user, service, process and dataflow, the model must be supplemented with layered analysis and explicit trust-boundary mapping. For security teams, the practical takeaway is to stop treating role clarity as a given and make it a design control.

A few things that frame the scale:

A question worth separating out:

Q: How should security teams handle delegated access when AI agents act on behalf of customers?

A: Security teams should treat delegated access as a separate governance layer, not as a normal login session. Define what the agent can do, how much value it can move, which approvals are required, and how delegation is revoked. Without those boundaries, the agent inherits more authority than the customer intended and fraud risk expands quickly.

👉 Read our full editorial: AI agents break fixed-role threat models: STRIDE vs MAESTRO



   
ReplyQuote
Share:

Free weekly newsletter

Subscribe to the NHI & AI Identity Journal

The latest on NHI and Agentic AI security – articles, research, breaches, news and events every week.

Bonus 33% off our NHI Course when you subscribe.