Case 79BA0F · AI security · L3 Applied

Adversarial Examples (Evasion)

Practise as: Explain itLevel: L3 Applied
  1. 01 What is it?
  2. 02 How is it abused?
  3. 03 How do you stop it?
What a strong answer covers

Try it out loud first. Then check yourself:

  1. Inputs perturbed, often imperceptibly, so a model misclassifies at inference; crafted with gradients (FGSM, PGD) or queries.
  2. Stakes: evading malware and phishing classifiers, stickers that fool vision models, spoofed face recognition.
  3. Black-box works too: perturbations often transfer between models, and score-based queries can estimate gradients.
  4. Defences: adversarial training, input transforms, ensembles; many published defences fell to adaptive attacks.
  5. Never make one model the only control: layer detections, limit score exposure, and watch for probing query patterns.

If the interviewer pushes back

  • Why does returning confidence scores from a malware classifier API make evasion easier, and what would you return instead?
  • How do evasion attacks on a classifier relate to jailbreaks on an LLM?

Go deeper

More AI security cards

All AI security cards →

cardosec draws a security topic and gives you a clock: explain it out loud with no notes, then see what you covered and what you missed. Free during early access.

Reviewed 2026-10-09