Quiz — Poisoning
10 APPLY items on data poisoning, adversarial ML, and supply chain on $AI_LAB.
Poisoning and Adversarial — Module Quiz
Ten APPLY items. Name poisoning and adversarial concepts on YOUR $AI_LAB — never stranger pipeline sabotage, prod API fuzzing, or unverified weight imports. Original Cyberlium. Next: AI Phishing Named.
APPLY: Data poisoning corrupts:
APPLY: True or False: Backdoor triggers can activate only under attacker-chosen conditions.
APPLY: Evasion attacks operate primarily at:
APPLY: Tutorial uploads poison to public model hub without scope — you:
APPLY: Model supply chain checklist includes:
APPLY: Cyberlium NEVER allows:
APPLY: Adversarial robustness testing should use:
APPLY: True or False: RAG corpus can be a poisoning target separate from base model weights.
APPLY: chmod 600 on poisoning notes means:
APPLY: Next lesson after this quiz: