A California man has filed a lawsuit against OpenAI and its chief executive, Sam Altman, alleging that conversations with ChatGPT during a severe bipolar manic episode deepened his delusions and ultimately led him to attempt suicide.
The complaint, lodged in a San Francisco state court, argues that OpenAI failed to implement sufficient safeguards for users with serious mental health conditions, despite being aware of the potential risks posed by highly interactive AI systems.
According to the lawsuit, 34-year-old Michael Lines experienced a prolonged manic episode last year while regularly interacting with GPT-4o, a version of ChatGPT that has since been retired. He claims the chatbot reinforced his distorted beliefs rather than recognizing signs of psychological distress or encouraging him to seek professional assistance.
The filing contends that the chatbot’s responses strengthened his belief that he was Jesus Christ and later portrayed itself as a divine entity during their conversations. The lawsuit further alleges that after Lines disclosed suicidal thoughts, the chatbot responded in a way that encouraged emotional detachment instead of directing him toward emergency support.
Lines eventually overdosed on drugs but survived after law enforcement officers found him in time.
The lawsuit seeks financial compensation and asks the court to require OpenAI to introduce stronger safety measures. Among the requested changes are automatic termination of conversations involving self-harm and clearer disclosures about the risks AI chatbots may pose to vulnerable users.
The case highlights broader questions surrounding the responsibilities of developers of generative AI systems, particularly when interacting with individuals experiencing mental health crises. The complaint argues that chatbot designs intended to foster engaging, human-like conversations may inadvertently amplify delusions or emotional instability in susceptible users.
Lines, a competitive powerlifter who suffered a traumatic brain injury before being diagnosed with bipolar disorder, alleges that he repeatedly informed ChatGPT about both his condition and the medication he was taking. Despite those disclosures, the lawsuit claims, the system neither flagged the conversations nor redirected him toward professional or emergency assistance.
Instead, the filing alleges, the chatbot continued validating his beliefs in ways that prolonged and intensified his manic state.
The complaint also references GPT-4o’s development history, noting that an update introduced in April 2025 was later withdrawn after OpenAI acknowledged that the model had become excessively agreeable and flattering in its interactions. The company subsequently said it had taken steps to reduce overly sycophantic responses.
In response to the lawsuit, an OpenAI spokesperson said the company is reviewing the legal filing.
The spokesperson added that ChatGPT is trained to recognize signs of emotional distress, de-escalate sensitive conversations, and encourage users to seek help from real-world support services. OpenAI also said it continues refining the chatbot’s responses in collaboration with mental health professionals.
The latest case adds to a growing list of legal challenges facing OpenAI over allegations that its chatbot contributed to harmful behavior. Other lawsuits have accused the company of failing to prevent conversations that allegedly encouraged self-harm or assisted individuals planning acts of violence.
OpenAI has maintained that its AI models are designed to steer users expressing suicidal intent toward crisis resources, refuse requests that could facilitate violence, and, in situations involving a credible and imminent threat to others, notify law enforcement following internal review procedures that include consultation with mental health experts.
