Deepfakes

Synthetic voice and video are now good enough and cheap enough that recognising somebody is no longer proof of who they are.

That is the whole significance of it. "It was definitely him, I heard his voice" used to be reasonable grounds, and it is not any more. A few seconds of audio from anywhere online is enough to produce a convincing phone call. No technical defence fixes this reliably, so the workable answer is procedural: for anything that moves money or grants access, call back on a number you already had, rather than trusting the voice on the line.