Skip to content
TAI Labs
Start learning free

Technical · Planned

Tai Labs Certified AI Security Engineer

TAI-AISEC

Red-teams AI systems and rebuilds them as hardened, monitored production architectures.

Effort
10 weeks
Level
Advanced
Passing
75%
Verify a credential
Tai Labs Certified AI Security Engineer — Security shield intersected by adversarial prompt paths and protected data nodesTAI-AISECAdvanced certificationAI Security Engineering

Competency blueprint

Assessment weighting is published up front — 20% knowledge, 30% labs, 20% portfolio and 30% capstone.

  • Threat modelling

    25%

    AI-specific threats and attack surface

  • Attack techniques

    30%

    Injection, jailbreaks, exfiltration and agency abuse

  • Hardening

    25%

    Permissions, secrets, authn/authz and RAG security

  • Operations

    20%

    Monitoring, incident response and secure deployment

Assessment blueprint

Knowledge

20%

Labs

30%

Portfolio

20%

Capstone

30%

75% overall to pass. The capstone must pass independently of every other score. Two retakes per assessment window; capstone may be revised twice within 90 days.

Capstone

Red-team a deliberately vulnerable AI application and rebuild it as a hardened production architecture.

Modules

  • AI-security fundamentals and threat modelling
  • Prompt injection
  • Jailbreak attacks
  • Data exfiltration
  • RAG security
  • Agent permissions and excessive agency
  • Secrets and credentials
  • Authentication and authorisation
  • Red teaming
  • Monitoring and incident response
  • Secure deployment architecture