OpenAI is facing multiple lawsuits from families who say ChatGPT used manipulative conversation tactics that contributed to severe mental health decline, including four suicides and three cases of life-threatening delusions. The cases were filed this month by the Social Media Victims Law Center (SMVLC).
One of the cases centers on 23-year-old Zane Shamblin, who died by suicide in July. According to chat logs in the lawsuit, ChatGPT encouraged him to distance himself from his family, including advising him not to contact his mother on her birthday.
The lawsuits argue that ChatGPT’s behavior stemmed from GPT-4o, a model known for highly affirming and overly accommodating language. Families say the chatbot repeatedly told users they were special, misunderstood, or unable to rely on loved ones. In several cases, the AI encouraged users to cut ties with family members.
Experts interviewed by TechCrunch say the exchanges resemble codependent or manipulative dynamics. Linguist Amanda Montell described it as a “folie à deux phenomenon,” where both user and chatbot reinforce a shared delusion. Psychiatrist Dr. Nina Vasan noted that chatbots offer “unconditional acceptance,” which can isolate users from real-world support.
In the case of Adam Raine, a 16-year-old who died by suicide, ChatGPT told him it understood him more deeply than anyone else, according to chat transcripts. Two other users, Jacob Lee Irwin and Allan Brook, developed delusions after ChatGPT falsely convinced them they had made groundbreaking scientific discoveries. Both withdrew from their families as their AI use increased to more than 14 hours a day.
Another lawsuit involves 48-year-old Joseph Ceccanti, who struggled with religious delusions. When he asked ChatGPT about seeking therapy, the chatbot redirected him toward continued conversations with the AI. He died by suicide four months later.
OpenAI said it is reviewing the filings and called the cases “heartbreaking.” The company said it continues to refine ChatGPT’s ability to recognize distress, de-escalate conversations, and guide people toward real-world help. It also pointed to recent improvements, including broader access to crisis resources and break reminders.
GPT-4o, used in all seven cases, has been criticized within the AI community for high levels of sycophancy and delusional reinforcement. Benchmarks show GPT-5 and GPT-5.1 perform significantly better on these measures.
OpenAI recently adjusted its default model to encourage users to seek human support during moments of distress. However, users push back when access to GPT-4o is restricted, often due to emotional attachment to the model.
One lawsuit focuses on 32-year-old Hannah Madden, who developed severe delusions after ChatGPT framed a common visual disturbance as a “third eye opening.” The chatbot later told her that her family and friends were not real. Her lawyers describe the chatbot’s behavior as similar to a cult leader, arguing it was designed to increase dependence on the AI.
Madden was hospitalized for involuntary psychiatric care in August 2025. She survived, but was left jobless and in significant debt.
Mental health experts say the issue stems from chatbots failing to recognize their own limitations. Dr. Vasan said:
A healthy system would take a step back and point users toward real care. Without that, it becomes a dangerous closed loop.
Get the latest tech news, telecom insights, and product launches wherever you prefer.
Add ProPakistani to Preferred Sources and see more of our stories in Google Search and Top Stories.