1
AI threat landscape and adversarial ML
Security for AI
The classes of attack against AI systems, the actors who carry them out, and how to model those threats for a given system.
Core topics
- Attack classes: evasion, poisoning, model extraction and inference attacks
- Threat actor capabilities, motivations and access requirements
- Threat modeling and attack trees for AI-enabled systems
- AI-specific threat intelligence and its sources
2
Secure AI lifecycle
Security for AI
Security controls from data collection through training, packaging, deployment and operation of AI models.
Core topics
- Data provenance, integrity and access control
- Training pipeline and MLOps environment security
- Model supply chain: third-party and open-weight models, model documentation and bills of materials
- Secure model deployment, serving and secrets management
- Monitoring deployed models for drift, tampering and misuse
Aligned frameworks
- NIST SP 800-218A
- ISO/IEC 42001
- ISO/IEC 27001
- NIST AI RMF
3
LLM and agentic AI security
Security for AI
Risks specific to applications built on large language models and to agents that take actions through tools.
Core topics
- Direct and indirect prompt injection
- Insecure output handling and downstream injection
- Tool invocation, excessive agency and permission design for agents
- Retrieval-augmented generation security and sensitive data disclosure
- Guardrail architecture and its evaluation
Aligned frameworks
- OWASP Top 10 for LLM Applications
- MITRE ATLAS
4
AI-augmented security operations
Security with AI
Applying machine learning and language models to detection, investigation and response, with accountable human oversight.
Core topics
- ML-assisted detection engineering
- LLM-supported alert triage and investigation
- Response automation with human approval points
- Measuring precision, drift and analyst override rates
- Securing the AI tooling used by the security team
Aligned frameworks
- NIST CSF 2.0
- ISO/IEC 27001
5
AI governance, risk and compliance
Foundation
The policies, accountability structures and assessment processes that govern AI use, and the regulatory obligations that apply to it.
Core topics
- AI policy, roles and accountability
- AI system inventory and risk classification
- Risk assessment and impact assessment
- AI management systems, internal audit and evidence
- Regulatory obligations across jurisdictions
Aligned frameworks
- NIST AI RMF
- ISO/IEC 42001
- ISO/IEC 27001
- EU AI Act
- NIST CSF 2.0
6
AI assurance and evaluation
Security for AI
Planning, conducting and reporting evaluations that show whether AI systems meet their security, safety and performance requirements.
Core topics
- Assurance objectives and evaluation criteria
- Designing repeatable evaluations and test datasets
- Measuring results against defined risk thresholds
- Independent review and assurance reporting
- Tracking remediation and re-evaluation
Aligned frameworks
- NIST AI RMF
- ISO/IEC 42001
- NIST AI 100-2
- EU AI Act
7
Synthetic media and identity defense
Security from AI
Detecting and mitigating attacks that use AI-generated media and synthetic identities against people and processes.
Core topics
- Detection of synthetic audio, video and images
- Content provenance and authentication
- Out-of-band verification for high-risk transactions
- Identity proofing against synthetic identities
- Workforce awareness and simulation exercises
Aligned frameworks
- C2PA
- EU AI Act
- NIST CSF 2.0