Loading demo links...
What to study for each D-AIS-F-A-00 domain
Introduction to AI Security
This domain introduces the AI threat landscape, including how AI systems, models, and data differ from traditional applications. It covers core attacks such as prompt injection, data poisoning, model extraction, evasion, and privacy attacks, plus AI assets, shadow AI, and supply chain risk.
Key skills
AI threat landscape, prompt injection and jailbreaking, data poisoning and backdoors, model extraction, evasion and privacy attacks, AI assets and supply chain risk, threat modeling
Reduce the Attack Surface (i.e., identifying threats and protecting application)
This domain covers hardening AI applications by reducing exposed endpoints, features, data, and privileges. It includes least-privilege tool use, input and output validation, trusted data and model sourcing, API protection, segmentation, and controls for retrieval systems and plugins.
Key skills
Attack surface reduction, least privilege for AI tools, input and output validation, data and model provenance, API hardening, network segmentation, retrieval and plugin security
Detect and Respond to Cyber Threats
This domain covers monitoring AI systems through logging of prompts, outputs, model versions, and tool calls, and detecting abuse such as extraction, injection, and jailbreak attempts. It also covers triage, containment, AI red teaming, threat hunting, and incident response procedures specific to models and data.
Key skills
AI logging and telemetry, anomaly and extraction detection, drift monitoring, incident triage and containment, red teaming, threat hunting, AI incident response planning
Recover from the Cyberattack
This domain covers restoring AI services after an attack, including rolling back or retraining models with clean data, quarantining poisoned data, and restoring prompts, configurations, and retrieval indexes from known-good backups. It also covers credential rotation, downstream remediation, validation before return to production, and lessons learned.
Key skills
Model rollback and clean retraining, data quarantine and revalidation, versioned and immutable backups, credential rotation, downstream remediation, recovery testing and post-incident improvement