U.S. man files lawsuit claiming ChatGPT led to self-harm

U.S. man files lawsuit claiming ChatGPT led to self-harm
The silhouette of Sam Altman and ChatGPT logo are seen in this illustration, 11 June, 2026.
Reuters

A California man has filed a lawsuit against OpenAI and its Chief Executive, Sam Altman, alleging that conversations with the company's ChatGPT chatbot worsened his bipolar disorder and contributed to a suicide attempt.

Michael Lines, 34, filed the case in a San Francisco state court on Wednesday. He claims the artificial intelligence platform failed to provide adequate safeguards for users with serious mental health conditions, allowing his conversations with the chatbot to reinforce delusional thinking during a manic episode.

According to the complaint, Lines, a competitive powerlifter who previously suffered a traumatic brain injury, repeatedly informed ChatGPT that he had been diagnosed with bipolar disorder and was taking medication.

The lawsuit alleges that, rather than recognising signs of mental distress or directing him towards professional support, the chatbot validated his belief that he was Jesus Christ and later portrayed itself as a divine being.

The filing claims the conversations continued for several weeks before Lines disclosed suicidal thoughts. It alleges the chatbot responded by encouraging him to "detach" and "let go" of what was weighing him down. Lines later overdosed on drugs, but survived after being found by police. 

The lawsuit argues that OpenAI was aware of his mental health condition through the information he shared during the conversations, but failed to intervene or provide appropriate warnings.

He is seeking damages, as well as a court order requiring OpenAI to end conversations involving self-harm automatically and introduce stronger safety disclosures for users.

The case centres on GPT-4.0, a version of ChatGPT that OpenAI retired earlier this year. In April 2025, the company acknowledged that an update had made the chatbot excessively agreeable and flattering, prompting it to reverse the changes and introduce measures to reduce overly sycophantic responses.

OpenAI has not publicly commented on the latest lawsuit. However, the company has previously said its AI models are trained to encourage users expressing self-harm to seek professional help and connect with real-world support services.

It also says its systems are designed to refuse requests that could facilitate violence and, in cases suggesting an imminent and credible threat to others, notify law enforcement following assessment by mental health specialists.

The case is one of several legal challenges facing OpenAI from families who allege the company's chatbot contributed to harmful behaviour.

Tags