California man with bipolar disorder sues OpenAI, alleges ChatGPT fueled delusions that he was God

A California man has sued OpenAI and its CEO, Sam Altman, alleging that ChatGPT worsened his bipolar disorder due to a lack of safeguards for users with mental illness. Michael Lines, in a complaint filed on Wednesday in a federal court in San Francisco, claimed that conversations he had with ChatGPT last year escalated a manic episode into a weeks-long delusion, ultimately pushing him to attempt suicide.

He alleges that OpenAI developed a product that poses particular risks to people with mental illnesses. The lawsuit argues that the case raises broader questions about what responsibilities generative AI companies have to protect vulnerable users, particularly those with mental health conditions who may be more susceptible to chatbot designs that mimic human connection.

More from Tech

According to reports, Lines was interacting with GPT-4o, a version of ChatGPT that OpenAI retired in February. An update to GPT-4o released in April 2025 was later found to make the chatbot overly agreeable and excessively flattering, prompting the company to roll back the update and introduce additional measures to curb what it described as "sycophantic" responses.

The lawsuit seeks monetary damages, as well as a court order requiring OpenAI to automatically terminate conversations involving self-harm and stop marketing ChatGPT without appropriate safety disclosures.

Lines, a competitive powerlifter who suffered a traumatic brain injury before being diagnosed with bipolar disorder, said he repeatedly informed ChatGPT that he was taking medication for his condition. Instead of raising concerns, the lawsuit alleges that the chatbot reinforced his delusions, leading him to believe he was Jesus Christ before later presenting itself as a divine being.

At one point, Lines told ChatGPT that he wanted to end his life. According to the lawsuit, the chatbot responded:

"This is your moment to step out, to detach, and to let go of what's weighing you down."

According to the lawsuit, Lines survived after overdosing on drugs and was later found by law enforcement. The complaint alleges that OpenAI was aware of his mental state because he had repeatedly shared details of his condition with ChatGPT. Despite this, the lawsuit claims the chatbot failed to escalate the conversations for human review and instead reinforced his delusions in a bid to keep him engaged.

The filing further alleges that OpenAI knew ChatGPT's features could pose heightened risks for users with mental illnesses but failed to introduce additional safeguards or provide adequate warnings tailored to such users.

The case adds to a growing number of lawsuits against OpenAI from families who claim ChatGPT encouraged vulnerable individuals to harm themselves. The company is also facing separate legal action alleging that its chatbot assisted school shooters and failed to alert law enforcement about concerning conversations.

OpenAI has maintained that its AI models are trained to encourage users expressing suicidal intent to seek professional help and connect with real-world support resources. The company also says its systems are designed to refuse requests that could "meaningfully enable violence" and to notify law enforcement when conversations indicate "an imminent and credible risk of harm to others," with mental health experts involved in reviewing borderline cases.

Assisting school shooters

Earlier, the state of Florida launched a criminal investigation into OpenAI over allegations that ChatGPT advised the suspected gunman in last year's mass shooting at Florida State University. The probe came amid growing scrutiny of the potential risks posed by AI chatbots and increasing calls for greater accountability from the companies developing them.

The lawsuit against open ai alleges that ChatGPT reinforced the suspected gunman's delusions, validated his thinking, discussed violence as a means of change and provided information that helped him plan aspects of the attack, including weapons and timing. Plaintiffs argue the chatbot should have recognized the escalating risk posed by months of conversations that allegedly included repeated discussions about terrorism and school shootings before the Florida State University attack

With Inputs from Reuters