💡Deepfake video calls are possible but still imperfect — watch for lip-sync delays, unnatural blinking, lighting inconsistencies, and reluctance to turn their head.
AI-generated video calls are the newest threat. Current deepfake technology can superimpose a face in real-time, but the technology has tells: lip movements slightly lag behind speech, blinking patterns can be unnatural, head turns cause momentary distortion, and lighting on the face may not match the background.
Tips for spotting deepfakes during a video call:
- Ask them to turn their head side to side — deepfakes struggle with profile angles
- Watch for lip-sync accuracy, especially with fast speech
- Note if their blinking seems regular and mechanical vs natural
- Ask them to hold an object near their face — deepfakes struggle with occlusion
- Check if the lighting on their face changes naturally when they move
This is an evolving area. The safest approach combines video calling with platform-level verification (like Smooch's ID + selfie matching through Yoti). Personal vigilance plus independent verification provides the strongest protection available.