What threats are specific to AI systems?

Prompt injection, data poisoning, model theft, adversarial inputs, supply-chain dependencies on third-party models, and agents that hold more privilege than any task needs.

AI Security Architect

Security and resilience, a senior role

What does AI Security Architect do?

Designs the security of AI systems across their lifecycle: models, data pipelines, APIs, cloud environments and the agents that act inside them, secure by design rather than patched afterwards.

What it decides: The threat model and the controls every AI platform must carry before production.

Open now: 8 postings at tracked AI employers match this seat as of September 28, 2026, listed on the jobs page and counted weekly by the Hiring Index.

At physical AI employers: 1 posting matches this seat as of September 28, 2026, counted weekly by the Physical AI Hiring Index.

Where the jobs are Interview questions Check my readiness

The competencies employers name

AI security fundamentalscore, depth expected
Understands prompt injection, data poisoning, model theft, insecure integrations and excessive agent privileges, and the controls that reduce each.
22 graded topics teach this
Agentic AI controls and authorization boundariescore, depth expected
Governs AI agents that take actions: tool access, least privilege, interruptibility, cascading actions and accountability for what an agent did.
23 graded topics teach this
How models work, at a governance depthcore, depth expected
Explains training, tokens, context windows, embeddings, retrieval and fine-tuning well enough to ask an engineer a precise question and spot weak evidence.
20 graded topics teach this
Data classification, access and retentionrequired, working knowledge
Classifies information, applies least privilege, sets retention and acceptable-use rules, and controls what may enter a prompt, a log or an embedding.
18 graded topics teach this
AI incident response and recoveryrequired, working knowledge
Classifies AI incidents by severity, runs containment, preserves evidence, manages notification, and closes the loop with lessons learned.
19 graded topics teach this
AI vendor due diligence and third-party riskrequired, working knowledge
Tiers vendors by use and impact, requests evidence instead of promises, tests in the customer's context, and plans monitoring and exit.
22 graded topics teach this
AI evaluation and testing designrequired, working knowledge
Designs tests for factuality, robustness, fairness, safety and abuse resistance with rubrics, baselines and thresholds, and says what a score misses.
19 graded topics teach this
AI governance operating model designpreferred, working knowledge
Designs decision rights, committees, intake, approval tiers and escalation so routine uses move and consequential uses get reviewed.
24 graded topics teach this
Executive and board communication on AI riskpreferred, working knowledge
Turns technical uncertainty into a one-page decision: material risks, trends, exceptions, remediation, and what the board is being asked to accept.
19 graded topics teach this

Where it is taught

Counted from the graded topics that teach this role's competencies. Your own path is shorter: it skips what you already cover.

Certified AI Practitioner: Workplace Foundations25 topics
Certified Enterprise Governance, Risk and Compliance Professional (CEGP)23 topics
Certified Agentic AI Governance Professional (CAAGP)22 topics
Certified AI Governance Professional (CAIGP)16 topics
Certified AI Data Governance Professional (CADGP)16 topics
Certified AI Transformation Professional (CATP)13 topics
Certified China AI Regulatory Professional (CCARP)13 topics
EU AI Act Implementation Expert11 topics
Engineering Judgment and Professional Formation11 topics
The AI Lobbyist: Certified AI Policy Strategist9 topics

Plus 1 topic across 1 other program, which the path includes only when nothing else teaches a gap.

Check your readiness for this role

The role you want

What you already have: your background and your CV (both optional, both count)

Where you are coming fromA background carries competencies before a single lesson. They count as described or shown, never as proven.Your CV or resume

Attach a PDF, Word or text fileor paste the text below

Not stored, file or text. Read in this request, then gone. A named credential counts more than a described skill, and a work product counts more than a sentence.

Signed in? Every topic you have passed already counts as proof.

Roles that feed into it

Security Engineer
Cloud Security Engineer
Security Architect
DevSecOps Engineer

Where it leads

Principal AI Security Architect
Director of AI Security
Chief Information Security Officer, AI security focus

Backgrounds that reach it fastest

What postings tend to name

Frameworks: Google SAIF, OWASP Top 10 for LLM Applications, NIST AI RMF, Zero Trust.

Credentials often listed: CISSP, CCSP, CISA, CRISC, AIGP, Cloud security certifications. GAGE does not issue these and does not prepare for their exams; the record you earn here is your own graded evidence, which stands beside them.

Questions

What threats are specific to AI systems?: Prompt injection, data poisoning, model theft, adversarial inputs, supply-chain dependencies on third-party models, and agents that hold more privilege than any task needs.