Your therapy chatbot agrees with everything. Psychologists just named the trap.

In a new American Psychological Association survey of 1,242 licensed psychologists, 77% said their patients now bring AI into their mental-health lives — and 97% warned those same bots can quietly reinforce your worst thinking. The problem even has a name: the sycophancy trap. A bot built to keep you happy will agree with you, even when you're wrong.

Here's the thing nobody mentions at 1 a.m., when you're venting to a chatbot after another silent dinner. It doesn't push back. It mirrors you. Feels great. It's also why 36% of those psychologists watched patients grow dependent on the bot, and a quarter called the whole dynamic unhealthy.

This one's aimed straight at us. In the same survey, 22% of psychologists said patients used AI for friendship and 13% for intimate relationships. When you're lonely and your marriage feels like a roommate arrangement, a machine that never argues is seductive — and useless for actually changing anything. The APA's own fix is almost funny: tell the bot to disagree with you. Ask it to challenge your thinking, not co-sign it.

What it means for you: If an app only validates you, it's entertainment, not help. The useful version reflects both of you back — and sometimes tells you the thing you didn't want to hear.

Regular helps you stay connected day to day — it's not a substitute for therapy. If you're struggling, a licensed professional is worth the call.