Security controls for AI systems
Protecting models, data, pipelines and AI interfaces.
Controls include access management for training data, models and APIs, protecting model weights, securing the training pipeline, logging prompts and outputs, input validation, rate limiting and secure handling of secrets used by AI agents.
On the exam
AI systems need normal security controls plus AI-specific ones; model files and training data are sensitive assets.
Related terms
- Model extraction and model inversionStealing a model, or recovering its training data, through its outputs.
- Privileged access management (PAM)Extra controls over powerful administrator accounts.
- AI agents (agentic AI)AI systems that plan and take actions using tools, not just produce text.
AAIA Test yourself on this with timed AAIA practice questions in the AAIA Exam Simulation Hub, or read the free AAIA exam guide.
Test yourself with today’s question of the day.
