Presentation attack detection, or PAD, is the set of checks used to tell a live person apart from a spoof presented to a sensor. The spoof may be a printed photo, a replayed video, a silicone mask, a virtual camera feed, or another fake input aimed at fooling a face, voice, or hand-based verifier.
PAD matters because many authentication systems trust what the sensor sees first. If a challenge can be satisfied by a still image or recorded motion, the system may grant access to a bot or imposter. Defenses include motion analysis, liveness cues, texture checks, depth signals, and timing tests that are hard to reproduce in a replay. In practice, PAD is usually one layer in a broader fraud stack, alongside rate limits, device checks, and server-side risk scoring.