Deepfake detection is the use of security controls, models, and verification workflows to identify AI-generated or manipulated audio and video. It looks for signs of synthetic media, such as unnatural facial motion, mismatched lip sync, audio artifacts, or inconsistencies in provenance metadata. Detection tools may run on endpoints, in cloud services, or inside collaboration platforms where users actually view the content.
It matters because deepfakes can be used for impersonation, social engineering, fraud, and executive-voice scams. A convincing fake video call or cloned voice can pressure staff into changing payments, approving access, or sharing sensitive data. In defense, detection should be treated as one signal, not proof of intent. Effective programs combine automated alerts with human review, identity verification, watermarking, content provenance, and clear incident-response rules. False positives can disrupt legitimate meetings, while false negatives can leave organizations exposed, so policy and user training are as important as the detector itself.