Phase 5 — AI Security / LLM Security / AI Governance

AI Security · LLM Security · AI Governance

Bảo mật trong kỷ nguyên AI: tấn công/phòng thủ ML, OWASP LLM Top 10, Prompt Injection nâng cao, quản trị AI theo NIST AI RMF / ISO 42001 / EU AI Act, và AI Red Teaming chuyên nghiệp. (Security for the AI era: ML attacks, LLM threats, GenAI defense, governance frameworks, red teaming.)

Danh sách chương — Chapter list

Nền Tảng Bảo Mật AI/ML

AI/ML Security Fundamentals — pipeline security, attack taxonomy (evasion/poisoning/extraction/inversion), adversarial robustness, MLOps security.

Mối Đe Dọa LLM & OWASP LLM Top 10

LLM Threats & OWASP LLM Top 10 — tất cả 10 rủi ro LLM, RAG security, vector DB poisoning, prompt injection detection, output validation.

Bảo Mật GenAI & Prompt Injection Nâng Cao

GenAI Security & Advanced Prompt Injection — jailbreaking, multi-modal attacks, agent security, NeMo Guardrails, Llama Guard, content safety.

Quản Trị AI — NIST AI RMF, ISO 42001, EU AI Act

AI Governance — NIST AI RMF 1.0, ISO/IEC 42001:2023, EU AI Act risk classification, AI Bill of Rights, algorithmic auditing, SHAP/LIME/Fairlearn.

AI Red Teaming & Adversarial ML

AI Red Teaming — PyRIT, MITRE ATLAS, Garak, adversarial defenses (randomized smoothing, feature squeezing), supply chain attacks on AI models.