Artificial intelligence models can now clone a voice with just a few seconds of audio, fueling a surge of deepfake songs ...