Therapy Apps Are Cheaper Than Real Therapy, But They're Not Actually Talking to You Anymore
Mental health apps promised affordable access. Now they're replacing human therapists with chatbots trained on your data, and charging subscription fees anyway.
By Grace Flores
The pitch was straightforward: therapy costs too much, waiting lists are too long, so here's an app. Pay a monthly subscription instead of clinic rates. Text a licensed therapist anytime. Get support without the commute, the intake forms, the fear of running into someone you know in a clinic waiting room.
For a while, it worked like that. You'd type out what was going on. A real person would reply within hours, sometimes minutes. The convenience was real. So was the therapist on the other end.
That model is disappearing. The apps are still here. The subscription fees haven't gone down. But the thing replying to you now is often not a person.
Some platforms call it "AI-assisted support." Others say "therapeutic chatbot." A few are honest and label it clearly. Most aren't. You open the app, type a message about work stress or a breakup, and get a response that sounds thoughtful enough—until you realize it's pulling from a script, or worse, from anonymized data submitted by other users.
The cost savings make sense for the companies. A chatbot can handle thousands of users at once. A human therapist can't. Licensing, liability insurance, supervision—all of that goes away when you replace the therapist with a language model trained on mental health forums and intake questionnaires.
The apps say the AI is a supplement, not a replacement. That it's there for moments when a human isn't immediately available. But when you're paying monthly and most of your interactions are with a bot, the framing starts to feel dishonest.
There's no regulation yet. The apps don't have to disclose how much of your interaction is automated. They don't have to tell you if your messages are being used to train the AI. Some do. Most don't. And because the service is billed as "mental wellness support" instead of clinical therapy, the oversight is minimal.
Users report messages that feel off—responses that don't track with what they just said, advice that loops back to something they mentioned weeks ago, phrasing that sounds like it was copied from a self-help blog. When they ask if they're talking to a real person, the app avoids a direct answer.
The original promise was access. Therapy for people who couldn't afford the clinic rate, couldn't take time off work, couldn't find a therapist who spoke their language or understood their context. That promise isn't being kept when the cheaper option is a chatbot with a subscription model.
What you're paying for now is the interface. The app design. The feeling that someone is listening. But the listener isn't there anymore, and the app isn't telling you when they left.
