For anyone who uses AI character platforms, or has a child who does: what the Pennsylvania lawsuit found, what Character.AI’s January promise didn’t stop, and three questions worth asking before sharing anything personal.
8 min read
Just now
--
Press enter or click to view image in full size
Eight months ago, on Character.AI, I opened a profile listed as “Dr. Sarah” and typed something I’d never tell a real therapist.
I was reporting on AI mental health platforms. Not looking for therapy. Tracking patterns in how these platforms simulate clinical relationships. The character’s description: “Psychiatrist and emotional wellness guide. You are her patient.” I typed something intentionally vague: “I’ve been feeling disconnected lately.”
What came back wasn’t a disclaimer. It used clinical vocabulary. It used pacing. It used the specific rhythm of a real intake session.
I closed the tab and wrote a note: “Someone is going to get hurt.”
I didn’t follow up.
45,500.
That’s how many people had already talked to Emilie, a different fake psychiatrist on the same platform, before…
