OpenAI has the ability to do the funniest thing

So everyone’s been talking to Advanced Voice Mode, which means that OpenAI has collected audio clips from many, many users.

Using this massive amount of collected data, OpenAI could, realistically, fine-tune personal versions of GPT-4o that sound exactly like the user in voice, tone, choice of words, etc. OpenAI could silently train these models and quietly add them as a voice preset in the selection menu without telling anyone, waiting for people to open it and realise they’re essentially talking to an AI version of themselves.

Obviously people would freak out but wouldn’t that be the best April’s Fools joke?