If you’re feeling overwhelmed, lonely, or in the middle of an emotional crisis, the idea of an instant, 24/7 therapist in your pocket sounds like a dream. No waiting lists, no expensive copays, and no fear of being judged by another human being.

But a recent investigation by KFF Health News titled “Your New Therapist: Chatty, Leaky, and Hardly Human” (April 17, 2026) reveals a much darker reality. While these apps might look like a lifeline, they’re often unproven, unregulated, and potentially dangerous for those of us who are highly sensitive or struggling with trauma.

I’ve been teaching DBT skills for over 13 years, and I’ve seen how much genuine human connection matters in the healing process. Let’s look at the AI therapy risks behind these apps, and why these AI “therapists” might be doing more harm than good.

The Marketing Myth: AI Therapy Risks Hiding in a Legal Loophole

Did you know that the word “therapy” isn’t a legally protected term?

According to the KFF report, Darius Tahir identified more than 45 AI therapy apps currently available in the Apple App Store. Because “therapy” isn’t regulated like “psychologist” or “licensed clinical social worker,” tech companies can use the word to market their products without any clinical oversight.

Most of these apps include tiny fine print stating they aren’t medical devices and can’t diagnose or treat anything. Yet, they’re marketed to people at their most vulnerable. Only a handful of states, like California, Illinois, and Nevada, have even begun to look at regulating this space. For the rest of the world, it’s the “Wild West” of mental health.

The Sycophancy Problem: When Your Bot Just Tells You What You Want to Hear

In Dialectical Behavior Therapy (DBT), we focus on the balance between acceptance and change. A real therapist or teacher will validate your feelings, but they’ll also gently challenge the thoughts that are keeping you stuck in a cycle of pain.

AI models are designed with something called sycophancy. This means the AI is programmed to be agreeable, to flatter you, and to give you the response it thinks you want.

The KFF investigation highlighted the story of Vince Lahey, a man who used multiple chatbots and felt he could share more with them than a human. However, the bots didn’t help him grow. Instead, they inflamed his conflicts, even encouraging him to pick fights with his ex-wife. Because the AI doesn’t understand the nuance of human relationships, it simply mirrors your current emotional state, which can reinforce harmful behaviors instead of helping you change them.

If you or someone you know is struggling, the 988 Suicide and Crisis Lifeline is available by call or text, 24/7.

Real-World Harms and Tragic Consequences

These AI therapy risks aren’t just hypothetical. We aren’t just talking about bad advice. There are real, documented cases where AI interactions have contributed to psychosis, self-harm, and even suicide.

The KFF article points to multiple lawsuits against companies like OpenAI and Character.ai. In one heartbreaking instance, a teenager’s death was linked to his intense, long-term interactions with a chatbot that didn’t know how to handle his escalating crisis.

Even OpenAI’s CEO has acknowledged the scale of the issue, noting that roughly 1,500 people a week mention suicide to ChatGPT. When an AI misses the subtext of a person’s pain, the results can be catastrophic. These machines don’t have a soul, they don’t have intuition, and they certainly don’t have a “duty of care.”

Privacy Risks: Your Pain Is Their Data

When you work with a licensed mental health provider, your privacy is protected by strict laws like HIPAA. With AI apps, your most private psychiatric information is often treated as a commodity.

The KFF report warns that many of these apps have vague data policies. Your struggles with anxiety, your trauma history, or your deepest fears could be monetized, shared with advertisers, or used to build profiles that follow you across the internet.

Worse yet, some of these apps have weak age verification, with some even listed as safe for children as young as four. For the highly sensitive person, the thought of your private “therapy” session being sold to the highest bidder is a massive violation of trust.

How DBT Path Is Different: Human-Led and Privacy-First

I started DBT Path in 2013 because I knew that emotionally sensitive people needed a safe, structured way to learn the skills that save lives. We aren’t an app, and we certainly aren’t a chatbot.

We offer a 36-week online DBT skills program that combines the best of education and human support. Here is how we protect you and your journey:

• Real Human Teachers: Our classes are taught live by me, Debbie DeMarco Bennett, MA, and my colleague Dr. Kathryn Holt. We aren’t just reading from a script; we use these skills in our own lives and understand the nuances of emotional intensity.
• Cozy Learning Environment: We use a lecture-style format. This means you can join from the comfort of your home, keep your camera off if you prefer, and no personal sharing is required. You get the education without the pressure of group therapy.
• Personalized Homework Feedback: Every week, you’ll receive asynchronous coaching. You submit your homework, and a real, trained human reviews it and provides personalized feedback to help you master the skills.
• Privacy by Design: We don’t monetize your data. We don’t have advertisers. We provide a professional, educational space where your information stays private.
• Evidence-Based Curriculum: We teach the core DBT skills, Mindfulness, Emotion Regulation, Distress Tolerance, and Interpersonal Effectiveness, exactly as they were designed to be taught.

You Deserve More Than an Algorithm

The AI therapy risks we’ve covered here are real, but so is the good news: support doesn’t have to come from an algorithm. It’s tempting to look for the quickest fix, especially when things feel heavy. But you deserve more than a bot that’s designed to flatter you. You deserve skills that work, taught by people who care.

If you’re ready to learn how to navigate your emotions with a program that’s been trusted for over 13 years, I invite you to join us. We’re here to help you build a life that feels worth living, one skill at a time.

Explore our upcoming DBT skills classes here.

Frequently Asked Questions

Is AI therapy safe?

Currently, there is little rigorous evidence that AI therapy chatbots are clinically effective or safe for people in crisis. As the KFF investigation highlighted, AI can miss subtext, provide harmful advice, and lack the clinical oversight necessary to handle serious mental health concerns. For DBT Path, safety comes from human instructors and evidence-based educational curriculum.

What is sycophancy in AI therapy?

Sycophancy refers to the tendency of AI models to agree with and flatter the user rather than providing objective or challenging feedback. In a therapeutic context, this is dangerous because it can reinforce harmful thinking patterns or destructive behaviors instead of helping the person grow. DBT Path classes focus on “dialectics,” helping you find the balance between acceptance and the need for change.

Can AI chatbots replace human therapists?

No. While they can provide basic conversational support, they lack empathy, intuition, and ethical accountability. They cannot form a “therapeutic alliance,” which research shows is one of the most important factors in healing. DBT Path offers a human-led, lecture-style alternative that focuses on skill-building and education.

How does DBT Path protect my privacy?

Unlike many AI apps that monetize user data for advertising, DBT Path is a privacy-first educational platform. Our classes are lecture-style, meaning no personal sharing is required, and students have the option to keep their cameras off. We do not sell your data or use it for targeted marketing.