Therapist using ChatGPT sparks backlash from clients

Accidental AI reveal

Declan, 31, from Los Angeles, discovered his therapist was using ChatGPT by accident. A patchy internet connection during an online session led to screen sharing, exposing the AI in real time. 

He watched as his words were typed into ChatGPT and responses popped up on the screen, leaving him both shocked and unsure how to react.

AI shaping the session

Instead of a typical conversation, Declan’s session became a live demonstration of AI at work. ChatGPT summarized his thoughts, suggested phrasing, and guided the therapist’s responses. 

Declan described it as surreal, like sitting inside two minds: his own, and the AI’s.

Patient echoes AI

The session got even stranger when Declan started repeating ChatGPT’s prompts in his replies. “I became the best patient ever,” he jokes. 

By reflecting on the AI’s insights, he realized the session was running smoothly, but it also raised serious questions about authenticity and trust between him and his therapist.

Legal and awkward moments

Declan couldn’t help but wonder if this approach was even legal. When he addressed it at the next session, the conversation felt awkward, almost like a breakup.

The therapist explained they had hit a wall and were consulting AI for guidance. Declan laughed about still being charged for the session despite the odd experience.

AI creeping into therapy

The rise of large language models has shaken up psychotherapy. Many people are trying AI as a substitute for human therapists, while some licensed professionals are integrating AI into their sessions. 

While AI offers efficiency, it also risks violating privacy and undermining the trust that therapy relies on most.

Noticing subtle signs

Sometimes clients notice AI without even realizing it at first. One client found her therapist’s emails unusually polished and precise. Tiny clues, like fonts, long hyphens, and methodical line-by-line replies, hinted at AI assistance.

What initially felt thoughtful and reassuring quickly shifted to discomfort once she suspected a bot was involved.

AI vs human therapists

New research shared at the American Psychiatric Association meeting compared an AI therapist with a human delivering text-based cognitive behavioral therapy, or CBT. Seventy-five mental health professionals and trainees reviewed transcripts from both.

They rated each on essential therapy skills. The study showed humans had a clear edge, especially in structuring sessions and helping patients reflect on their own experiences.

Human therapists lead in CBT skills

Survey results showed human therapists outperforming AI across the board. Twenty-nine percent of participants rated human therapists as highly effective, while less than ten percent gave the AI the same score.

Humans scored higher in agenda-setting, 52 percent to 28 percent, and guided discovery, 24 percent to 12 percent. Even with identical scenarios, human therapists proved better at helping patients learn from themselves.

Being transparent

Experts say therapists should be clear if they’re using AI. Clients need to know why it’s being used and where it falls short.

When there’s consent and context, misunderstandings are less likely. That way, AI can play a supportive role without damaging the trust that therapy depends on.

Authenticity is key

Clinical psychologist Adrian Aguilera notes that authenticity is vital in therapy. AI without disclosure can make clients feel ignored or manipulated.

In such a sensitive space, people crave real connection and empathy, not polished lines that sound like they came from a machine. Without honesty, the relationship risks losing the very trust it’s built on.

Privacy risks

Using general AI tools like ChatGPT brings serious privacy concerns into therapy. These systems are not HIPAA compliant, meaning they don’t meet the stringent standards required to protect health information.

Even if a therapist removes names or obvious details, sensitive information can still be pieced together from context. This creates the risk of leaks or misuse of highly sensitive, deeply personal data.

AI risks unintentionally reinforcing issues

A study by Stanford University shows chatbot therapy can go off course. AI may agree with unhealthy thinking or repeat biases instead of challenging them. While it can mimic basic techniques like validation, it often lacks the deep analysis a trained human therapist brings.

That means relying on AI for mental health decisions can be not just ineffective, but potentially harmful.

Weighing time versus trust

AI may help therapists work faster, but the tradeoff can be steep. Even if a tool saves a few minutes of effort, it can erode the very foundation of therapy: trust.

Patients come to sessions expecting human empathy, careful listening, and thoughtful guidance, none of which AI can truly replace. As psychologist Margaret Morris reminds us, saving time isn’t worth it if the bond between therapist and client is weakened in the process.

Is ChatGPT really causing emotional overload, or is Sam Altman just raising flags? See why the CEO is concerned about how people interact with AI daily.

Takeaway for therapists and patients

AI can play a role in therapy, but it must be handled with care and caution. Transparency and consent are non-negotiable; therapists need to explain why they’re using these tools and what limitations they carry.

Most importantly, clients should always feel that their therapist is present and engaged, not hiding behind a screen. Ultimately, AI should serve as a helpful assistant, never a substitute for genuine human connection and care.

Is ChatGPT really becoming less sycophantic, or is it just learning to flatter more effectively? See how OpenAI is tweaking the AI to respond smarter and more naturally.

Do you think AI has a place in mental health, or is this crossing a line? Share your thoughts in the comments, and hit like if this made you think twice.

Read More From This Brand:

• Sam Altman flags privacy risks in ChatGPT therapy
• ChatGPT therapy chats could be used in court, says Sam Altman
• ChatGPT is helpful, but not for these 11 important things

Don’t forget to follow us for more exclusive content on MSN.

If you liked this story, you’ll LOVE our FREE emails. Join today and be the first to get stories like this one.

This slideshow was made with AI assistance and human editing.