Clarification - it’s already been built. OpenAI is trying to decide whether or not to unleash Voice Engine on the masses. They obviously realize the dangers, and offer suggestions on how to “bolster societal resilience against the challenges brought by ever more convincing generative models”.
Ugghh. I suppose if a company develops a robot predator that hunts humans, the least they could do is give us some tips on how to bolster our resilience before they release it…