What employers and buyers are asking for

We researched public 2026 sources on AI-security and AI-governance skills, then matched them against what the CAIN-42 codebase can genuinely teach. Each certificate exists because both sides lined up.

CAIN Certified Agentic Security Practitioner Very high

AI skills now appear in more than a quarter of G7 cybersecurity job postings, and hands-on experience with AI agents is the hardest entry-level competency to find. The OWASP Top 10 for Agentic Applications gave the field a shared vocabulary in December 2025.

CAIN MCP Security Engineer Very high

MCP became the default way agents reach tools. Dedicated MCP security courses and a hands-on MCP certification launched in 2026, and dozens of CVEs were filed against MCP systems early in the year.

CAIN AI Insurance & Risk Underwriting Analyst High (emerging)

Insurers are moving AI spend into underwriting and risk selection, report a widening talent gap in underwriting, and treat AI liability as an emerging risk with little loss history. Underwriters need evidence they can read.

CAIN AI Governance & Compliance Evidence Specialist Very high

Searches for AI governance certification more than doubled year on year in the UK, AIGP and ISO/IEC 42001 auditor roles are growing, and buyers increasingly ask suppliers for evidence of AI controls.

CAIN Agentic Red Team Operator High

AI red-team certifications (OSAI+, CAISP, COASP) multiplied in 2026 and OWASP is cataloguing agentic red-teaming tools. Organisations deploying agents need people who can attack them safely and turn findings into fixes.

CAIN Byzantine Trust Fabric Engineer Specialist

A niche but valuable skill: few teams can build and operate Byzantine fault-tolerant notarization. It underpins tamper-proof audit logs for regulated AI.

CAIN Evidence & Forensics Verifier High

As AI decisions face audits, disputes and incident response, organisations need people who can verify machine evidence independently instead of trusting a vendor dashboard.

CAIN Governed Memory & Data Protection Specialist High (emerging)

Memory and context poisoning is now a named agentic risk, and every agent that remembers or retrieves creates a data-protection question: what was stored, who can see it, and can it be erased?

CAIN Agent Assurance & Reliability Engineer High

Teams shipping agents need CI for agent behaviour: regression suites, trajectory grading and drift control. Vendors increasingly use LLM judges; deterministic grading is a scarce skill.

CAIN Formal Verification & Proof Engineer Specialist

AI-written code is multiplying, and proofs that cover every input are the strongest answer to 'how do you know it is correct?'. Few engineers can use SMT solvers in practice.

Sources (researched 2026-11-02)

Read with care: many certification rankings are written by training vendors, and job-market figures are often directional. The Cisco consortium data is the most independent signal listed.