A low-tech solution from the past may be your best defense against AI deepfakes

A ZDNet report describes how increasingly realistic AI-generated deepfakes have been used in social engineering attacks — including an Arup incident that led to about $25M in fraudulent wire transfers and a case where a training firm was infiltrated — and warns that voice/video alone is no longer reliable for authentication; it recommends low-tech mitigations such as hardware security keys, randomized verbal passphrases, out-of-band callbacks, and dual authorization to reduce risk.