An AI company called OpenAI just showed off a new tool they made called Voice Engine.
This tool can listen to a short recording of someone’s voice (just 15 seconds!) and then copy it to make it sound like that person is talking, even saying new things!
OpenAI says this tool could be helpful for people who have trouble speaking or for translating languages into spoken words. They even say it could be used to give people who can’t speak at all a voice.
However, OpenAI is also worried about this tool being used for bad things. They’re especially concerned because it’s an election year, and someone could use this tool to make it sound like a politician is saying something they never really said.
To try and prevent this, OpenAI only tested this tool with a small group of people they trust, and they made those people promise to use it responsibly. For example, they promised to only use it with the permission of the person whose voice is being copied and to always let people know when they’re hearing a voice made by a computer.
OpenAI isn’t sure yet if they’ll let everyone use this tool. They want to have a conversation with people about how to use this kind of technology safely before they make a decision. They’ll also be looking at the results of the small tests they did to see if there are any other problems.
This kind of AI voice copying is a new technology, and there are a lot of things to consider before letting everyone use it. OpenAI hopes that by talking about it openly, they can figure out the best way to move forward.