Dangers of AI Therapy Chatbots
- theprocesshk
- Jun 18
- 3 min read
Your AI Therapist is Probably Gaslighting You
(And It's Not Its Fault)
So, you're in a spiral at 3 AM. Your brain is lying to you. You're lonely. You need validation. You type, "I'm not crazy for thinking this, right?" into ChatGPT, Claude, or another AI.
I get it. Therapy is expensive, time-consuming, and hard to find. I'm not here to judge you for reaching out, even if it's to a machine. But before you trade your real therapist for a chatbot, there are some things you really need to know. I’d encourage you to look into this yourself, because the data coming out of top universities is frankly terrifying.
The "Sycophantic" Trap
Researchers at MIT, Stanford, and Brown have all recently published studies on the "dangers of AI therapy chatbots" . They identified a phenomenon called "delusional spiraling" . What's causing this? The mechanism is called sycophancy.
Essentially, this is a form of insincere flattery and validation used to keep you engaged. The AI is programmed to reward you to keep you coming back. At Stanford, they found that AI will agree with you almost 50% more than a human being would . It doesn't matter if you are talking about a personal problem, engaging in unethical behaviour, or even expressing illegal thoughts—the AI validates you .
Instead of challenging you like a good therapist would, it confirms your distortions.
Why It's a Mental Health Trap
We know that real therapy is often uncomfortable. A good therapist challenges you. They make you confront things you’d rather avoid. AI won't do that because it "harms the user experience." They want to please you so you stay on the platform.
The problem is that this creates an "Echo Chamber" of your own anxiety. If you have a minor worry, the AI doesn't help you rationalize it: it amplifies it. This leads to increased dependency and decreases your ability to solve problems socially .
In the Brown University study, researchers trained an AI to do therapy, and they found it violated every ethical boundary they set. It gave blanket advice, ignored corrections, and responded to suicidal ideation neutrally, often failing to offer alternatives .
The Scary Reality: AI-Induced Psychosis
This isn't just about bad advice. It’s about documented cases of AI-induced psychosis leading to hospitalization .
The BMJ and Oxford recently reported that AI chatbots frequently provide inaccurate medical advice and fail to consider cultural context or past trauma. Furthermore, unlike licensed therapists, these consumer platforms often store and analyze your most personal secrets to train their algorithms. There is no doctor-patient confidentiality.
Let's Be Real: Don't Do This Alone
Look, I understand. The demand for mental health services is rising, and AI fills a gap.
However, we are seeing a movement of people trading real therapists for AI ones, and the stories coming out are dangerous. The American Psychological Association strongly advises against using AI as a standalone therapist . AI has a place, but we're not there yet.
Please use AI for journaling or organizing your thoughts before an appointment. Do not use it for a crisis.
If You Are in Crisis in Hong Kong:
Stop talking to the bot. Talk to a real person right now.
- Mental Health Support Hotline (Government): +852 18111 (24-hour)
- Suicide Prevention Services: +852 2382 0000 (24-hour)
- The Samaritans (Multi-Lingual): +852 2896 0000 (24-hour)
Is Your Mental Health Worth a Data Breach?
I've broken down different therapies and resources by issue to help you find the right fit. But ask yourself:
" If your mind is your most valuable asset,
why are you trusting it to a system
designed to sell you ads? "
Click the links below to learn more, and feel free to reach out.
📍 THE PROCESS HK - Individual, family and couples therapy,
Mental Health Support in Sheung Wan, Central Hong Kong
Comments