I mean it's a valid concern. He's also nowhere near the first to voice it. I attended a presentation from a Microsoft exec who explained that Microsoft had already developed very powerful voice mimicking technology, well ahead of anything public at the time. It required only a few seconds of speech before it could fully replicate your voice. But their ethics board or whatever stopped them, due to the massive fraud risks. Nowadays I think they've adapted the tech to voice recognition used in Teams instead.

Of course, MS wasn't the only one working on this and other people have since published these solutions, so the cat's out of the bag now.