Let's face it, high costs, limited time, and even then, among taboos seeking help for mental health, it's no wonder that more and more people are embracing chatbots like ChatGpt as “therapists.”
However, AI has given advice and made tamed ears more accessible, but also comes with the very real risk of reports of AI psychosis, hospitalizations and even suicide.
Still lying on the ChatGpt sofa and want to share all the issues? This post talked to clinical psychologists about how to do it safely and the key to making the most of it.
“As a clinical psychologist, I don't think ChatGpt is a treatment alternative. It has the nuances, attachment needs and emotional dynamics that require human connection and harmony,” says Clinical psychologist Ingrid Clayton PhD. And the author of the book “Fawning” posted it.
But that doesn't mean you can't use it at all. Many of her own clients use AI across sessions in useful ways, as long as the technology is displayed and implemented as a supplement rather than an alternative.
“For example, clients can run dating app messages or emotionally charged text via AI to get neutral feedback and help them recognize patterns of emotional inability to use, bias, manipulation,” she said.
“It was amazing to learn that these insights reflect what we have already explored in the session.”
Other clients are looking for a nervous system regulation tool that can be used in real-time using AI at moments of dysregulation.
“It's not treatment, but it can help support the treatment process and bridge insights and skill building between sessions,” she added.
In Clayton's case, there is an inherent risk to relying solely on AI for treatments, including lack of personalization. Because your bot doesn't know your history, trauma, or context, “that response can miss or misinterpret important emotional nuances, especially when our own blind spots shape the questions we ask.”
Read Clayton's tips on how to make the most of AI for therapeutic support.
1. Use it as a tool, not as a replacement
Clayton said that AI should be used in tandem rather than traditional therapists.
2. Please ask for specific, actionable instructions
Specificity is important and skepticism is required.
“Please ask specific and included questions,” urged Clayton. “You get the most useful response by seeking something viable that will help you reconstruct the movements and messages that you're grounding, rather than seeking wide range of emotional guidance.”
Researchers have found that bots are perfect for people. This is because humans prefer to be confirmed in a match than correction, leading to users being more positively rated.
Surprisingly, a 2025 survey found that popular therapy bots like Serena and “Therapist” and 7 cups answered just about half of the prompts properly.
3. Beware of emotional dependence
AI can provide false sense of security to people struggling with mental health issues. “The ability to mimic empathy and therapeutic language can lead users to believe they are receiving professional care when they are not,” Clayton said.
She urges users to be wary of the tendency to consistently use AI for emotional support, including daily validation and decision-making.
“Excessive dependence can promote self-renunciation, an outsourcing of your inner knowledge to external (and unrelated) sources. For people with relational trauma, this can reinforce the pattern of doing the 'right' rather than doing it right on their own,” she said.
4. Keep notes for the therapist
“Reality checks advice with experts. If something resonates deeply or feels anxious, explore therapist more fully and in context,” Clayton added.
In this way, AI can become a hot topic rather than absolute.
5. Know the limits of crisis
Clayton emphasized that bots should not be dependent on life-threatening situations because they are not equipped to deal with suicidal ideation, abuse or acute trauma.
“At that moment, reach out to a licensed therapist, a trusted supporter, or a line of crisis,” she said.
In fact, a 2025 Stanford University survey found that people experiencing delusions, suicidal ideations, hallucinations, and OCD were inappropriate and dangerous statements for people experiencing at least 20% of LLM.
