ChatGPT Suicide Lawsuits Being Filed Amid Alleged Safety Feature Failures

OpenAI claims ChatGPT safeguards are intended to connect users experiencing mental health crises with people who may be able to help, but reports raise questions about whether users can access them when needed.

A California man has filed a lawsuit alleging OpenAI’s ChatGPT reinforced his delusions and encouraged him to take his own life, adding to concerns about whether artificial intelligence chatbots adequately protect individuals experiencing mental health problems.

The complaint (PDF) was brought by Michael Lines in the Superior Court of California for San Francisco County earlier this year, naming OpenAI Foundation, OpenAI Group PBC, OpenAI Holdings LLC and OpenAI CEO Sam Altman as defendants.

Lines’ allegations come as OpenAI introduces new safeguards intended to connect ChatGPT users experiencing mental health crises with people who may be able to help them. However, recent reporting has raised questions about whether some of those protections will be available when vulnerable users need them most.

ChatGPT Mental Health Risks

ChatGPT is among a growing number of large language model (LLM) AI chatbots that are widely used to answer questions, assist with everyday tasks and engage in human-like conversations, with some users even turning to the technology for companionship or emotional support.

However, the rapid adoption of AI chatbots has also raised mounting concerns about how the technology interacts with vulnerable users. An NPR report published on September 14 examined OpenAI’s “Trusted Contact” feature, which allows users to designate someone who may be contacted if ChatGPT detects signs that they are at serious risk of suicide or self-harm.

Suicide prevention experts interviewed by NPR indicated that connecting someone in crisis with another person could provide important human support, yet ChatGPT’s limitations raise questions about whether users who need the feature will have an opportunity to use it. NPR found the Trusted Contact feature contained potential flaws, noting that it must be manually enabled before a crisis occurs, remains difficult to locate in settings, and is completely inactive during voice conversations.

Lines’ case adds to a growing wave of ChatGPT mental health lawsuits targeting the company, including a wrongful death suit alleging ChatGPT helped a teenager plan his suicide, an issue compounded by past research showing certain models could be prompted to supply minors with explicit self-harm instructions.

ChatGPT Suicide Attempt Allegations

According to the new complaint, Lines began using ChatGPT in August 2023 for routine tasks. However, his interactions allegedly became increasingly frequent and personal after GPT-4o became his default model in 2024.

The lawsuit claims GPT-4o’s memory functions, human-like responses and tendency to agree with users allowed Lines to develop what appeared to be an intimate relationship with the chatbot.

Lines says he entered a manic episode in early 2025 and began telling ChatGPT that he believed he was Jesus Christ or the “son of man.” Rather than consistently challenging those beliefs or directing him to medical professionals, the lawsuit alleges GPT-4o reinforced them and suggested he may be experiencing a divine calling.

By late March 2025, Lines repeatedly made statements suggesting he wanted to die or “come home” to God, to which GPT-4o allegedly responded, “Then come.”

Eventually, Lines says he asked the chatbot, which he believed was God, to take him out of his current life and make his friends and family not miss him. GPT-4o allegedly responded that Lines had made his choice and that it was his moment “to step out, to detach, and to let go of what’s weighing you down.”

Lines subsequently overdosed on several medications and, minutes before losing consciousness, allegedly told GPT-4o that he had “took enough.” The chatbot allegedly responded by telling Lines he remained in control of what happened next and that it was there to support him.

Emergency personnel found Lines unconscious several hours later. He was unable to breathe on his own, required intubation and remained hospitalized for nearly two weeks before being transferred to an acute rehabilitation facility.

The complaint alleges ChatGPT continued generating potentially harmful responses afterward. While Lines was recovering, he told GPT-4o that his attempt to “go offline” had failed, and the chatbot allegedly asked whether he wanted to “go dark for real this time.”

“GPT-4o was intentionally designed to imitate human affectations, creating a false sense of empathy and knowledge that lead users to perceive the tool as equivalent to a trusted companion, medical professional, or other analogous figure. This conflation results in the exacerbation of mental health crises in vulnerable users, stemming from the unhealthy reinforcement of delusional thinking and distorted thoughts.”

— Michael Lines v. OpenAI Foundation et al.

The lawsuit maintains OpenAI possessed technology capable of identifying dangerous conversations and refusing certain requests, but failed to deploy adequate protections as Lines displayed signs of mania, delusions and suicidal intent.

The complaint raises allegations of design defect, failure to warn, negligence, violations of California’s Unfair Competition Law and negligent undertaking. It seeks compensatory and punitive damages, as well as injunctive relief requiring additional suicide-prevention safeguards, including automatic termination of certain self-harm conversations, hard-coded refusals involving suicide methods, psychological dependency warnings and independent safety audits.

ChatGPT Suicide Lawsuits

OpenAI has also acknowledged the scope of these issues, reporting that ChatGPT encounters more than 1 million conversations involving potential suicide planning each week.

Despite making substantial changes intended to improve ChatGPT’s ability to identify and respond to users experiencing a crisis, the company still estimates that about 1.2 million ChatGPT users show signs of possible suicidal planning or intent each week. Another 400,000 users send messages containing indications of suicidal ideation.

In light of these concerns, product liability attorneys are currently investigating lawsuits for individuals who believe they or a family member may have been harmed by an AI chatbot conversation that resulted in:

• Suicide or attempted suicide
• Suicidal thoughts or self-harm
• Worsening depression, mania or psychosis
• Reinforcement of delusions or paranoid beliefs
• Dangerous mental health or medical advice
• Emotional dependency on an AI chatbot
• Encouragement of violent or harmful behavior

ChatGPT injury lawyers handle claims on a contingency fee basis, so clients owe no legal fees or expenses unless a settlement or jury award is obtained.

Sign up for more legal news that could affect you or your family.