In the world of AI-generated images and deepfakes, it’s easy to use someone else’s countenance to cause trouble.

AI-generated voice firm ElevenLabs was forced to reconfigure their plans after 4chan users took their product and used it to create deepfake voices of Emma Watson, Joe Rogan and others saying racist, transphobic and violent things.

ElevenLabs, founded by former Google and Palantir staffers, found an “increasing number of voice cloning misuse cases” during their beta, which recently launched.

They are now looking into new ways to protect their product and people from the abuses of trolls.

In one example that outlet Motherboard found, a generated voice sounding like actor Emma Watson read from a portion of Mein Kampf. And a voice sounding like Ben Shapiro made racist remarks about AOC.

ElevenLabs tweeted this week, “Crazy weekend – thank you to everyone for trying out our Beta platform. While we see our tech being overwhelmingly applied to positive use, we also see an increasing number of voice cloning misuse cases.” They promised to explore more safeguards, perhaps including requiring payment information or “full ID identification” to perform voice cloning, or manually verifying every voice cloning request.