
The bipartisan GUARD Act would criminalize exploitative AI interactions with children and require companies to disclose when users are speaking to a machine.

Researchers posing as children on Character AI logged 669 harmful interactions in 50 hours, including grooming and drug offers.