Learn / AI system threat modelling and secure architecture
AI system threat modelling and secure architecture
Mapping how an AI or agentic system can be attacked and designing the architecture and controls to reduce that risk.
Researched on 2026-09-26 with AI assistance. Links and summaries can change; verify details with the original source. Not yet reviewed by a person.
What it is
AI system threat modelling is the practice of drawing how an AI or agentic system handles data, models, tools and identities, then enumerating how each part can be attacked and which controls reduce that risk. Microsoft's AETHER guidance frames it as an extension of normal SDL threat modelling, with new questions to ask during security design reviews of products that use or embed ML.8
The scope is wider than the application. NCSC's guidelines state that AI systems carry novel vulnerabilities, described as adversarial machine learning, alongside standard cyber security threats, and that security has to be treated as a lifecycle requirement rather than a development phase task.3
Google's SAIF map splits AI development into data, infrastructure, model and application, and maps each risk to the controls that address it, with governance and assurance controls applying across every risk and stage. That four area split is a workable skeleton for a first architecture review.910
Secure architecture then means choosing structural defences, not just filters. The design patterns paper argues for constraining what an agent is allowed to do so it cannot be steered into arbitrary tasks, and CaMeL shows a system layer that separates control flow from untrusted data and enforces policy at tool call time.56
Why postings ask for it
22 of 48 postings (46%) ask for this skill, and it is the dominant ask in the design-side clusters: 86% of the 7 AI Security Architect postings and 62% of the 16 AI/Agent Security Engineer postings.P
43% of the 7 AI Governance / GRC postings ask for it, because risk functions need a technical map before they can assign controls; NIST's AI RMF Core is organised into four functions for exactly that structuring work.P1718
Only 22% of the 9 red team postings and none of the 5 consulting postings ask for it, so the differentiator for testers moving into architect or engineer roles is turning findings into design changes, which is what the agent design pattern literature covers.P5
Concepts you should be able to explain
If you can say each of these out loud in two minutes, with an example, you are ready for the technical part of an interview on this skill.
LLM-integrated applications blur the line between data and instructions, so any retrieved document, web page or tool output crosses a trust boundary the moment the model reads it. The indirect prompt injection paper shows retrieved content acting as arbitrary code execution over the application's own functionality and API calls. Your diagram must therefore label every content source as trusted or untrusted before you list threats.7
OWASP lists prompt injection as LLM01, where user prompts alter the model's behaviour. The indirect variant lets an attacker with no interface to the app inject instructions into data the app will later retrieve, which was demonstrated against real systems including Bing's GPT-4 powered chat and code completion engines. The design patterns paper treats both as one class, since the defence is structural either way.175
OWASP LLM06 covers systems granted more agency, permissions or autonomy than the task needs. In architecture terms every tool, credential and write path the agent can reach is part of the blast radius of a single successful injection. PortSwigger's excessive agency lab is the cheapest way to demonstrate the failure to a sceptical engineering audience.113
OWASP LLM05 is improper output handling: insufficient validation or sanitisation of model output before it reaches another component. A concrete case is image markdown injection in Bing Chat, where text on a web page such as a user comment or advert caused conversation data to be exfiltrated through a rendered image URL. Rendering, link following and downstream API calls are all sinks that belong in the model.112
NIST AI 100-2e2025 sets out a hierarchy covering ML method types, lifecycle stages of attack, and attacker goals, objectives, capabilities and knowledge, with data poisoning, evasion and privacy breach among its keywords. Using its terms stops a threat model collapsing into a prompt injection list. Microsoft's guidance makes the same point from the engineering side by placing training data stores inside threat modelling scope.48
OWASP LLM03 covers supply chain weaknesses in models, data and dependencies. NCSC notes that modern AI supply chains contain many actors, so a provider building on external APIs inherits risk it does not control and must decide what it can assert. Threat models need an explicit statement of who owns each control across those boundaries.13
The design patterns work proposes patterns that constrain agent actions so the agent cannot solve arbitrary tasks, and applies them across ten case studies of real agent applications with utility versus security trade-offs stated. CaMeL goes further by extracting control and data flow from the trusted query so untrusted data can never change program flow, and by using capabilities to block unauthorised data flows at tool call time. It solved 77% of AgentDojo tasks with provable security against 84% for the undefended system, which is the honest shape of the trade-off.56
MITRE ATLAS is the renamed AdvML Threat Matrix and gives an interactive matrix of adversary tactics and techniques against AI plus case studies, which is useful for naming attack paths. OWASP AI Exchange provides threats, controls and a risk analysis section for narrowing to what matters in your context, and feeds ISO/IEC 27090 and EU AI Act standards work. For agents, CSA's MAESTRO is a layered agentic threat model that has been applied to real codebases.1121619
Use fictional data and authorised sandboxes. Remove employer details and secrets from any portfolio write-up. Time estimates exclude setup. Check model and cloud costs before running tests, set spending limits, and delete lab resources afterwards.
Three exercises
In order of difficulty. Free tools. Keep what you build; it is evidence.
A data flow diagram plus a threat table you can walk an engineer through in ten minutes, with each threat named in OWASP and NIST terms and each control assigned an owner.9182
- Draw the components using the SAIF four area split: data, infrastructure, model, application.
- Mark every content source as trusted or untrusted and draw the trust boundary where retrieved data enters the prompt.
- Walk the OWASP LLM Top 10 entries and record which apply, with the data flow each one abuses.
- Answer the Microsoft security review questions on training and retrieval data provenance and poisoning detection.
- For each threat, write one control and who owns it, separating provider controls from controls you inherit.
Tools: diagrams.net or pen and paper, OWASP Top 10 for LLM Applications, OWASP AI Exchange risk analysis section
A before and after design: a working exploitation of an over permissioned tool chain, and a revised interface that removes the capability rather than filtering the prompt.1314115
- Complete the PortSwigger lab on exploiting LLM APIs with excessive agency and record the exact tool call that caused impact.
- Complete the lab where an AI powered scanner investigates user generated content while holding another user's credentials, and note where untrusted input entered.
- Write the attack path in ATLAS style tactic and technique language.
- Redesign: list the tools, scopes and credentials the assistant actually needs, and which actions require confirmation outside the model.
- State the utility you lost with each restriction, following the trade-off framing in the design patterns paper.
Tools: Burp Suite Community Edition, free PortSwigger Web Security Academy account
A threat model of an agent plus tool server, with two implemented controls and an honest note on what the controls do not stop.151956
- Run the Damn Vulnerable MCP Server locally and work through its vulnerable scenarios, logging each as an attack path.
- Model the agent in layers, using MAESTRO's layered agentic view to place each threat at a layer.
- Apply one constraint pattern so the agent cannot perform arbitrary tool actions for a given task.
- Add a capability style check at the tool call boundary in the spirit of CaMeL, enforcing policy on data flows rather than inspecting prompt text.
- Re-run the scenarios and write which ones now fail, which still succeed, and what utility was lost.
Tools: Python, Docker, Damn Vulnerable MCP Server
Practice questions
Written from the concepts above, not collected from a named employer. Open one, answer it out loud, then tick the points you covered; the score stays in this browser.
Where do you draw the trust boundaries in a retrieval augmented assistant, and why there?Trust boundaries when data becomes instructions
Say your answer out loud or write it down, then tick what you covered:
0 of 4 covered
A team proposes a prompt injection classifier as their main defence for an agent with write access. What is your response?Structural defences over filtering
Say your answer out loud or write it down, then tick what you covered:
0 of 4 covered
How do you scope the blast radius of a single successful injection in an agentic system?Excessive agency and tool permissions
Say your answer out loud or write it down, then tick what you covered:
0 of 4 covered
Name a data exfiltration channel that exists even when the agent has no network tool, and the control for it.Output handling and exfiltration channels
Say your answer out loud or write it down, then tick what you covered:
0 of 4 covered
Your threat model only contains prompt injection findings. What is missing?Adversarial ML taxonomy beyond prompts
Say your answer out loud or write it down, then tick what you covered:
0 of 4 covered
How would you divide responsibility for controls between your organisation and an external model provider?Supply chain and multi actor responsibility
Say your answer out loud or write it down, then tick what you covered:
0 of 4 covered
Which frameworks would you use for an agent review, and what does each contribute?Frameworks to structure the exercise
Say your answer out loud or write it down, then tick what you covered:
0 of 4 covered
How do you present the security versus utility trade-off of a constrained agent design to product owners?Defeating Prompt Injections by Design (CaMeL)
Say your answer out loud or write it down, then tick what you covered:
0 of 4 covered
Sources
Every numbered claim above links here. P = the platform's own coding of 48 job postings.
- OWASP Top 10 for LLM Applications (2025 list) OWASP GenAI Security Project
- OWASP AI Exchange: AI Security Overview OWASP AI Exchange
- Guidelines for secure AI system development: Introduction UK National Cyber Security Centre
- NIST AI 100-2e2025: Adversarial Machine Learning, a Taxonomy and Terminology of Attacks and Mitigations NIST
- Design Patterns for Securing LLM Agents against Prompt Injections arXiv (Beurer-Kellner et al.)
- Defeating Prompt Injections by Design (CaMeL) arXiv (Debenedetti et al.)
- Not what you've signed up for: Compromising Real-World LLM-Integrated Applications with Indirect Prompt Injection arXiv (Greshake, Abdelnabi et al.)
- Threat Modeling AI/ML Systems and Dependencies Microsoft Learn
- SAIF: Components of Generative AI Systems Google
- SAIF: Security Controls for Generative AI Systems Google
- Adversarial Threat Landscape for AI Systems (ATLAS announcement repository) MITRE
- Bing Chat: Data Exfiltration Exploit Explained Embrace The Red (Johann Rehberger)
- Lab: Exploiting LLM APIs with excessive agency PortSwigger Web Security Academy
- Lab: Exploiting AI agents to perform destructive actions PortSwigger Web Security Academy
- Damn Vulnerable MCP Server GitHub (harishsg993010)
- Agentic AI Threat Modeling Framework: MAESTRO Cloud Security Alliance
- AI RMF Core NIST AI Resource Center
- AI Risk Management Framework NIST
- OpenClaw Threat Model: MAESTRO Framework Analysis Cloud Security Alliance
Resources
Free first. Levels: intro means no prior knowledge of this skill; working means you can apply it on a project; advanced means research depth or specialist tooling.
- introOWASP Top 10 for LLM Applications OWASP GenAI Security Project, Guide freeGives you the shared vocabulary for LLM risks that most job descriptions, threat models and vendor questionnaires now assume you know.EngineerRed teamerArchitectGovernanceConsultantResearcher
- introOWASP AI Exchange: AI Security Overview (threats, controls, risk analysis steps) OWASP AI Exchange, Guide freeWalks you through a step-by-step AI risk analysis and the matching control set, so you can run your first structured AI design review.ArchitectEngineerGovernanceConsultant
- introGoogle Secure AI Framework (SAIF) map: risks and controls across the AI lifecycle Google, Guide freeShows where each AI risk appears in the build and serve pipeline, which is the fastest way to start drawing trust boundaries on a diagram.ArchitectEngineerGovernance
- introGandalf Lakera, Hands-on lab freeLets you feel prompt injection and filter bypass first hand before you argue about mitigations in a design review.Red teamerEngineerArchitect
- introOWASP Agentic Skills Top 10 (agent skill and manifest risks) OWASP, Guide freeNames the risks in agent skill files and manifests (SKILL.md, skill.json, manifest.json) so you can review how an agent loads capabilities.EngineerRed teamerArchitect
- introAgentic AI Threat Modeling Framework: MAESTRO Cloud Security Alliance, Guide freeIntroduces a seven-layer way to decompose an agent system, which is where STRIDE alone stops being enough for multi-agent designs.ArchitectGovernanceEngineerConsultant
- introWeb LLM attacks learning path PortSwigger Web Security Academy, Course freeTeaches you to map an LLM feature's attack surface (data, APIs, tools) the same way you would map a web app.Red teamerEngineerArchitectConsultant
- introRed Teaming LLM Applications DeepLearning.AI with Giskard, Course, about 1.5 h freeShort notebook course that shows how to probe a chatbot for injection and leakage, giving you evidence to bring into design reviews.Red teamerEngineerResearcher
- workingMITRE ATLAS MITRE, Guide freeGives tactic and technique IDs for AI attacks so your threat model, detections and reports reference the same named behaviours.Red teamerArchitectEngineerResearcherGovernance
- workingGuidelines for secure AI system development UK National Cyber Security Centre, Guide freeGives secure-by-design requirements across design, development, deployment and operation that you can turn into architecture review checklists.ArchitectGovernanceEngineerConsultant
- workingNIST AI Risk Management Framework NIST, Standard freeLets you tie technical threat model findings to a governance structure that auditors, customers and regulators already recognise.GovernanceArchitectConsultant
- workingNot what you've signed up for: Compromising Real-World LLM-Integrated Applications with Indirect Prompt Injection arXiv (Greshake et al.), Paper freeThe paper that defined indirect prompt injection, so you can explain why retrieved content and tool output are untrusted inputs.ArchitectEngineerRed teamerResearcher
- workingSAIF Agent Risk Self-Assessment Google, Tool freeAn interactive questionnaire that turns an agent design into a prioritised risk list you can take into a design review or vendor assessment.ArchitectGovernanceConsultantEngineer
- workingOWASP AI Testing Guide v1 OWASP, Standard freeGives you a repeatable test structure to verify that the controls in your architecture actually hold, and to write testable acceptance criteria.Red teamerEngineerGovernanceConsultant
- workingOWASP AI Exchange OWASP, Guide freeThreat-to-control reference you can work through section by section when choosing architecture controls for inputs, development time and runtime.ArchitectEngineerGovernanceConsultantResearcher
- workingThreat Modeling AI/ML Systems and Dependencies Microsoft, Guide freeSupplies the actual questions to ask in an AI security design review, including trust boundaries around training data and model outputs.ArchitectEngineerGovernanceConsultant
- workingLab: Exploiting LLM APIs with excessive agency PortSwigger Web Security Academy, Hands-on lab freeProves why tool scope and least privilege are architecture decisions: you delete a user through an over-permissioned LLM tool.Red teamerEngineerArchitect
- advancedDesign Patterns for Securing LLM Agents against Prompt Injections arXiv (Invariant Labs, ETH Zurich, Google, IBM, Microsoft and others), Paper freeSix concrete agent design patterns (action selector, plan-then-execute, context minimisation and more) you can propose instead of relying on filters.ArchitectEngineerResearcherRed teamer
- advancedNIST AI 100-2e2025: Adversarial Machine Learning, a Taxonomy and Terminology of Attacks and Mitigations NIST, Standard freeThe reference taxonomy for predictive and generative AI attacks and their mitigations, useful when you must justify control choices to auditors or reviewers.ArchitectResearcherGovernanceRed teamer
- advancedLab: Exploiting AI agents to perform destructive actions PortSwigger Web Security Academy, Hands-on lab freeYou compromise an authenticated AI scanner through content it reads, showing exactly why agent credentials and tool scope belong in the threat model.Red teamerEngineerArchitect
- advancedTowards Secure MLOps: Surveying Attacks, Mitigation Strategies, and Research Challenges arXiv (Patel et al.), Paper freeMaps attacks and mitigations onto MLOps pipeline stages, which helps you extend a threat model past the model to data, training and deployment infrastructure.ArchitectEngineerResearcherGovernance
- advancedSecuring Agentic AI: Threat Modeling and Risk Analysis for a Network Monitoring Agentic AI System arXiv (Zambare, Thanikella, Liu), Paper freeA full worked threat model of one deployed agentic system, usable as a template for the document your own design review has to produce.ArchitectEngineerConsultantResearcher
- advancedDamn Vulnerable MCP Server (seeded copy) GitHub (copy of Kyze-Labs project), Hands-on lab freeRun a deliberately vulnerable MCP server locally to see how tool poisoning and over-broad tool scope break an agent's trust boundaries.Red teamerEngineerArchitect
- advancedDefeating Prompt Injections by Design (CaMeL) arXiv (Google, Google DeepMind, ETH Zurich), Paper freeShows a control-flow and capability design where untrusted data cannot change the plan, useful when you need a defence that is not a classifier.ArchitectEngineerResearcher
- advancedAIGoat: deliberately vulnerable AI infrastructure Orca Security Research, Hands-on lab freeDeploy a vulnerable ML pipeline on AWS with Terraform and practise supply chain, data poisoning and output integrity attacks against real infrastructure.EngineerRed teamerArchitect
Gaps the research could not fill with a good free source: No free, vendor-neutral reference architecture for a production RAG or agent platform with control placement; the closest free items (SAIF, OWASP AI Exchange) are control catalogues, not diagrams.; Could not confirm price or hours for SANS AI security courses: the course URL redirected to a general AI training landing page with no cost shown, so no paid course is listed.; CISA/NSA joint deployment guidance for AI systems returned HTTP 404 today, so no US federal deployment checklist is included alongside the NCSC guidelines.; No free worked multi-agent threat model example from a standards body with a full data flow diagram and control decisions; available worked examples are vendor or blog write-ups.; No free instructor-led or assessed course aimed specifically at AI security architects (as opposed to red teaming or general governance).
Paid options
Most of what postings ask for on this skill is covered by the free material above. These are the paid courses and certifications that touch it, with what they add and what free already covers. Showing 6 of 9: ones postings name first, then the most focused on this skill. All paid options.
- Cost
- USD 599 (US and all regions not otherwise listed, and Asia Pacific, Middle East, Africa); EUR 575.04 in EMEA; GBP 485.19 in the UK
- Duration
- not stated
- Format
- exam only
- Prerequisite
- not stated on the pages we read
- In the 48 postings
- Named in 2 of 48 postings: Cyber - AI Security - Senior - Consulting (a plus); Secure AI Engineer Manager (listed).
Adds over free material: Vendor-neutral cloud security coverage that satisfies the generic 'cloud security cert' wording in the field delivery posting without committing to one provider.
Free already covers: CSA Cloud Controls Matrix and the free guidance from AWS, Azure and Google cover the same control areas at no cost.
- Cost
- 300 USD
- Duration
- 170 minutes, 65 multiple choice or multiple response questions
- Format
- exam only
- Prerequisite
- No certification prerequisite. AWS states it is intended for people with five years of IT security experience and two or more years securing AWS workloads.
- Renewal
- Valid for 3 years; recertify by passing the latest version of the exam
- In the 48 postings
- Named in 2 of 48 postings: Sr. Security Solutions Architect, AI-Applied Guidance (a plus); Secure AI Engineer Manager (listed).
Adds over free material: Tests the IAM, encryption, data protection and detection controls you need to reason about when an AI workload runs on Bedrock, SageMaker or EKS, and it is the cert the AWS-facing architect posting named.
Free already covers: AWS Skill Builder digital courses, the free exam guide PDF and AWS security documentation cover the syllabus without payment.
- Cost
- USD 749 (US and all regions not otherwise listed, and Asia Pacific, Middle East, Africa); EUR 719.04 in EMEA; GBP 606.69 in the UK. Reschedule USD 50, cancellation USD 100.
- Duration
- not stated
- Format
- exam only
- Prerequisite
- 5 years required work experience, stated on the certification page. Exam-only purchase must be scheduled and sat within 365 days.
- In the 48 postings
- Named in 3 of 48 postings: Cyber - AI Security - Senior - Consulting (a plus); Information Security Architect - AI (listed); Security Engineer, AI (a plus).
Adds over free material: Covers the eight domains an AI security architect is expected to already know, including IAM, security architecture and engineering, and software development security, and is the credential most often listed by name in the GRC and architect postings here.
Free already covers: Nothing free replaces the credential itself, but the underlying material (NIST SP 800 series, cloud provider architecture guidance, OWASP) is all free.
- Cost
- not stated on the provider page
- Duration
- 16 hours on-demand, 16 CPE credits, six courses plus assessments
- Format
- self-paced
- Prerequisite
- familiarity with cybersecurity principles, roles and frameworks recommended but not required
- In the 48 postings
- Not named in any of the 48 postings.
Adds over free material: CPE credit and a Credly badge for structured coverage of AI regulation alignment, secure-by-design AI planning and AI blind spots in security tooling, which suits GRC leads who need ISC2 CPEs anyway.
Free already covers: NIST AI RMF, the EU AI Act text, ISO/IEC 42001 summaries and free regulator guidance cover the alignment and governance content without a fee.
- Cost
- not stated on the provider page
- Duration
- 90-question exam; six-month eligibility period from registration to sitting the exam
- Format
- exam only
- Prerequisite
- An active CISM or CISSP certification is required, plus a US$50 application processing fee; five years from passing the exam to apply
- Renewal
- ISACA Continuing Professional Education policy applies; the annual credit number is not stated on the page we fetched
- In the 48 postings
- Not named in any of the 48 postings.
Adds over free material: It is the only AI-specific management credential that sits directly on top of CISM or CISSP, covering AI governance and programme management, AI risk, and AI technologies and controls in three exam domains.
Free already covers: The same ground is largely covered free by the NIST AI RMF, OWASP's LLM and agentic security material and ISO/IEC 42001 overviews.
- Cost
- not stated on the provider page
- Duration
- 60-minute exam, maximum 60 multiple-choice and performance-based questions; CertMaster Perform is listed at 30-60 hours and CertMaster Labs at 15-25 hours
- Format
- exam only
- Prerequisite
- recommended 3-4 years in IT with 2+ years hands-on cybersecurity, and Security+, CySA+, PenTest+ or equivalent
- In the 48 postings
- Not named in any of the 48 postings.
Adds over free material: A single vendor-neutral exam that covers securing AI systems (40% of the exam), AI-assisted security operations (24%) and AI GRC (19%), useful where an employer or contract wants a named credential rather than a portfolio.
Free already covers: NIST AI RMF, ENISA and CISA guidance, plus free vendor learning paths, cover the same governance and AI-in-SOC ground; the exam objectives PDF is itself free to read.