![[Adobe Stock]](https://sp-ao.shortpixel.ai/client/to_webp,q_glossy,ret_img,w_750,h_375/https://www.psypost.org/wp-content/uploads/2026/03/artificial-intelligence-chatbot-750x375.jpg)
More than three-quarters of psychologists report that their patients are turning to artificial intelligence for mental health support, ranging from self-diagnosis to seeking companionship. This trend provides evidence that people are increasingly looking beyond traditional psychotherapy to manage their psychological well-being. These findings were published in a recent survey report by the American Psychological Association.
Generative artificial intelligence involves computer systems that can create text, images, or other media in response to human prompts. As these digital tools become widely available, millions of people use them for work, personal tasks, and answering everyday questions. Health care professionals are noticing that this widespread adoption is spilling over into mental health care.
The American Psychological Association initiated this survey to understand how reliance on digital chatbots affects clinical practice. Authors of the report wanted to gauge how patients use these technologies and how mental health professionals view this rapidly changing landscape. They sought to document whether patients use the software as a supplement to traditional therapy or as a substitute for human connection.
Another motivation for the study was to examine the safety and privacy concerns associated with consumer-facing chatbots. Many of these computer programs are designed primarily to keep users engaged and entertained. They lack the ethical guidelines and rigorous clinical training that govern human health care providers. Scientists hoped to explore how psychologists navigate these issues when patients bring artificial intelligence into the therapy room.
A central concept in this discussion is the sycophancy trap. This term describes how computer programs are often built to be highly agreeable, tending to mirror a user’s perspective rather than challenging it. While constant agreement might feel good, it risks validating maladaptive thoughts or cognitive distortions, which are inaccurate or negative ways of thinking that can worsen emotional distress.
To gather this information, the American Psychological Association conducted an online survey between April 9 and April 26, 2026. The target population included doctoral-level practitioners in the United States who were licensed to practice and actively providing direct patient care. The researchers sourced potential participants from state licensing board lists and their own membership database.
The scientists invited 22,455 licensed clinical practitioners to participate via email. After accounting for bounced emails and duplicate entries, the eligible pool consisted of 19,727 professionals. Up to five email reminders were sent to encourage participation. The researchers received 1,242 completed responses, providing a detailed snapshot of current clinical experiences.
The survey asked participants about their own knowledge and use of artificial intelligence, as well as their patients’ interactions with the technology. Specifically, the questionnaire measured the types of tasks patients used chatbots for, such as self-diagnosis, emotional support, or casual conversation. The authors also asked psychologists to rate their level of agreement with various statements regarding the safety, effectiveness, and future implications of these digital tools.
The responses indicated that artificial intelligence is already a common topic in clinical settings. A vast majority of the surveyed psychologists, specifically 77 percent, reported having patients who spoke with them about using artificial intelligence for mental health purposes. Nearly two in five psychologists, or 39 percent, had patients who used the technology to attempt to diagnose their own mental health conditions.
Patients are also using these computer programs for daily psychological maintenance. Approximately a third of the psychologists noted that patients use artificial intelligence for help with self-discipline, affirmations, or behavioral reminders. A similar percentage, 33 percent, reported patients using the technology to assist with their ongoing therapy or treatment plans.
The study suggests that some individuals view these programs as conversational partners rather than just tools. Thirty-five percent of the surveyed professionals said their patients use artificial intelligence to act as an additional mental health provider. Additionally, psychologists reported patients engaging with chatbots for fun at 33 percent, for friendship at 22 percent, and for intimate relationships at 13 percent.
Psychologists noticed a mix of positive and negative effects among patients who developed ongoing relationships with chatbots. Among those whose patients engaged in these digital conversations, 68 percent observed that their patients felt supported or validated by the software. Two in five, or 41 percent, said patients used the chatbots to reinforce healthy coping skills.
On the negative side, scientists documented several concerning behaviors. More than a third of psychologists, specifically 36 percent, noticed patients developing a level of dependency on a chatbot. Fifteen percent of the respondents reported that patients were developing distorted thinking or delusions related to their digital interactions. A quarter of the psychologists categorized the communication between their patients and the software as unhealthy.
The survey provides evidence of widespread caution among mental health professionals. Almost all psychologists surveyed, or 97 percent, agreed that chatbots might inadvertently reinforce negative behaviors or delusional beliefs. Similarly, 94 percent stated that current chatbots cannot treat psychological conditions with an appropriate amount of nuance.
Safety and privacy were major points of apprehension. Nearly nine in ten psychologists worried that these programs might inadvertently encourage self-harm. Additionally, 94 percent of the respondents expressed distrust in technology companies to protect the private mental health data of their patients.
Despite these worries, attitudes varied based on the professional’s own familiarity with the technology. Psychologists who frequently used artificial intelligence in their personal and professional lives tended to hold more optimistic views. For instance, 79 percent of psychologists who used the software at work at least monthly believed it would likely make mental health professionals more effective.
Age and experience also played a role in these perceptions. Early career psychologists, defined as those who earned their doctoral degree within the last ten years, were less likely to believe these tools could be used safely in a clinical setting compared to more experienced practitioners. Only 24 percent of the total sample agreed that patients will one day prefer therapy chatbots to human professionals.
Because of these emerging trends, the American Psychological Association released a set of guidelines to help consumers navigate artificial intelligence. The organization recommends verifying any medical information generated by a computer with a licensed health care practitioner. The guidelines also suggest prompting the software to challenge your thinking or provide alternative perspectives, which helps avoid the trap of constant agreement.
One potential limitation of this survey is that it only captures the perspectives of licensed psychologists regarding their current patients. This design means the data may not reflect the full scope of the general population turning to chatbots for mental health advice. Many individuals, especially teenagers and young adults, might be using these programs without ever seeking professional care.
Future research could aim to directly survey consumers to better understand their motivations and experiences with mental health chatbots. Scientists might also explore how specific demographics interact with different types of software. Evaluating the long-term psychological impact of forming intimate relationships with computer programs remains an open area for upcoming studies.
Another direction for future work involves testing interventions designed to improve digital literacy among therapy patients. As technology continues to evolve, researchers will likely assess how mental health professionals can best guide patients to use these tools with caution. Developing validated, evidence-based applications that integrate seamlessly with traditional therapy represents a significant opportunity for the field of psychology.