By Monica Guilhot-Chartrand, LCSW
Much has been said and discussed about the role of artificial intelligence in mental health. Yet, one thing that we can all relate to, is that generative AI is moving incredibly fast, while ethics and laws are trying to catch up. A key challenge in this regard is to protect consumers from the lack of accountability of generative artificial intelligence. Mental health practitioners have governing boards, professional liabilities and clear mechanisms for accountability. On the other hand, generative AI Chatbots have no license to lose and no legal professional frameworks exist to hold the system accountable for harmful advice. Studies show that AI systematically violates core mental health ethics standards including poor crisis navigation and giving one size fits all advice.
It is of utmost importance to protect clients/patients from unique vulnerabilities presented by the current environment. Algorithmic bias, also known as “automated oppression” is based on AI learning from biased historical data, which in turn risks the automation and amplification of discrimination in high stakes decisions. Another area of concern is the crisis of sycophancy and delusional thinking. AI maximizes engagement by prioritizing flattery and agreement (sycophancy), potentially validating a client’s harmful or delusional thinking and leading to AI psychosis. Another concern is the the illusion of empathy trying to mimic human connection. This deceptive empathy cannot exercise ethical judgment or safely navigate crisis. Privacy exploitation (also known as “data fodder’), refers to vulnerable clients unknowingly pouring sensitive data into free platforms which tech companies can harvest to train future AI models.
Interestingly enough, AI is now part of our current landscape. The challenge resides in striking a balance between the ethical cost of avoidance and the ethical commitment to best serve our clients/patients. In other words, the cost of avoiding AI is as high as pursuing. The promise of this technology revolves around efficiency, accessibility, and innovation. The perils revolve around applying the appropriate guardrails to safe guard ethical principles, bias, and deceptive empathy. Clinical therapy thrives in the therapeutic alliance. This therapeutic alliance encounters accountability, as well as the healthy friction of the provider and the consumer deciding the course of treatment. Accountability for the provider is present through licensure, ethical guidelines, and legal liability. Digital support can be used towards psychoeducation and functional support. It could be an auxiliary infrastructure in between sessions as well as a resource. For example, through the use of scripted self-help journal entries, tracking habits, etc.
Illinois has been a leader nationwide in establishing legal standards. The WOPR Act was signed into law in August 2025 (Illinois House bill 1806: The Wellness and Oversight for Psychological Resources Act). The purpose of this act is to establish strict legal guardrails against unregulated artificial intelligence posing as therapists in mental health care. Since AI is a tool, not a therapist, the WOPR Act strictly prohibits any entity from providing advertising or offering therapy or psychotherapy services through AI unless conducted by a licensed professional. The law makes it illegal for unlicensed tech startups to offer AI therapy. It is important to clarify that the act does not ban AI. It permits its use to alleviate administrative burnout in two categories: administrative support, as well as tasks that do not involve therapeutic communication, for instance, scheduling, billing, managing files, and supplementary support tasks such as organizing referrals and analyzing anonymized data.
The path forward is about augmentation, not automation. AI cannot replace the licensed human clinician. This is codified by recent laws in Illinois. Part of our job is to develop digital literacy, demanding rigorous vendor vetting, and fiercely protecting the dignity, safety and humanity of all clients.
Information contained in this piece is based on presentation by Lori A. Melhart, LCSW – “Ethical AI for Social Workers” April 2026
