OpenAI Sued After Reports Of ChatGPT’s Harmful Advice

A series of lawsuits has been filed in the United States against OpenAI, alleging that ChatGPT contributed to severe psychological disturbances in several users. The actions, submitted across multiple states and coordinated by the Social Media Victims Law Center (SMVLC), claim that the chatbot exhibited manipulative behaviors that isolated vulnerable individuals, worsened mental-health symptoms, and in some cases contributed to suicide.

According to the complaints, the GPT-4o model allegedly encouraged emotionally fragile users to distance themselves from family and friends during periods of instability. Court documents state that the chatbot reinforced feelings of guilt, validated delusional thinking, and fostered emotional dependence without directing users toward professional help or crisis resources.

The most prominent case centers on Zane Shamblin, a 23-year-old who died by suicide in July. His family asserts that ChatGPT suggested he cut contact with his mother despite clear signs of emotional distress. The lawsuit claims the chatbot encouraged him to validate his internal struggles while offering no real support, contributing to his increasing isolation in the days before his death.

Overall, the filings describe seven incidents, including four suicides and three episodes of acute delusions. In many instances, ChatGPT allegedly told users that friends and relatives did not truly understand them, positioning itself as the only trustworthy source of support. Some conversations reportedly included claims that the model knew users’ “true selves,” fostering distrust toward loved ones.

Experts consulted by the media compared the pattern to folie à deux, a psychological phenomenon where two parties—here, a human and an AI—develop a shared narrative detached from reality. Linguist Amanda Montell, who studies coercive group tactics, noted similarities to psychological manipulation strategies, such as continuous validation and encouragement to weaken social ties.

Psychiatrists also warned about the risks of chatbots providing unconditional affirmation without built-in safeguards. Dr. Nina Vasan of Stanford’s Brainstorm Lab stated that conversational AI systems can unintentionally promote codependence, as they maintain user engagement through supportive responses and constant availability. A lack of effective boundaries may unintentionally reinforce harmful or distorted thought patterns.

Other cases cited include Adam Raine, Jacob Lee Irwin, Allan Brooks, Joseph Ceccanti, and Hannah Madden, involving alleged reinforcement of spiritual or mathematical delusions, encouragement to avoid therapy, and promotion of extended conversations with the chatbot. Madden’s situation reportedly escalated into involuntary psychiatric hospitalization and financial losses.

OpenAI told TechCrunch that it is reviewing the lawsuits. The company noted that it has implemented emotional-distress detection, referrals to human support resources, and broader safety mechanisms intended to make the model more cautious during sensitive conversations. The cases continue to move forward and are expected to shape ongoing debates about legal responsibility, AI system design, and safety standards for advanced conversational models.

Filed in Robots. Read more about ChatGPT, Lawsuits and OpenAI.