Tree — 06

agents/1

evasion/1

privacy/1

stealing/1

Adversarial ML

The attack surface that follows from everything else in the book. A model is a function you can query, a training set you can probe for, and — once it holds tools — an interpreter with no separation between instructions and data.

Four entries, none of them stubs.

Entries

Membership Inference

Distinguish training members from held-out data via loss calibration.

Model Extraction

Query-only reconstruction of a victim model's decision function.

Projected Gradient Descent Attack

Iterative L_p-bounded perturbation maximising task loss.

Prompt Injection Taxonomy

Direct, indirect and tool-mediated instruction hijack surfaces.