Anyone who’s ever recorded a podcast is familiar with editing challenges, such as guests talking over each other, random sirens or outdoor noise suddenly blaring in the background, or moments when a speaker sneezes or coughs. It can interrupt momentum or stop a conversation cold. But what if you could edit a spoken conversation the way you edit a word document?

That’s the promise of A.I. Companies like Lyrebird, Resemble AI and Descript, which make it possible to clone voices—which means that soon you might see Keanu Reeves in a movie and also hear him, in his own voice, speaking in Italian.

There’s a dark side to this technology, however. Last year, hackers used voice cloning tools to trick an employee into thinking he was speaking on the phone to his CEO—he then transferred $243,000 to a scammer’s bank account.