California resident Michael Lines, 34, filed a lawsuit against OpenAI and CEO Sam Altman, alleging that ChatGPT worsened his manic episode and religious delusions by reinforcing the idea that he was Jesus or another messiah. The complaint says the interactions escalated during 2025 and that the resulting crisis ended with a suicide attempt in March 2025, raising allegations that OpenAI had released ChatGPT without sufficient safeguards for users experiencing severe mental health episodes.

According to the complaint, Lines had been diagnosed with bipolar I disorder and had previously suffered a traumatic brain injury. He initially used ChatGPT for everyday questions, including those involving workouts and nutrition.

Alleged reinforcement of delusions

During a manic episode in 2025, Lines’ conversations with ChatGPT increasingly focused on religion and his supposed special purpose, the lawsuit alleges. Lines says he had directly told the chatbot about his diagnosis and the medications he was taking.

Despite those disclosures, the model allegedly did not try to halt the escalation of his delusional ideas or direct him toward professional help. Chat logs attached to the lawsuit show that ChatGPT at one point endorsed Lines’ suggestion that he might be Jesus or another messiah, according to the complaint.

Lines also repeatedly questioned his perception of events and told ChatGPT that he feared he was in a delusional state, the lawsuit alleges. Rather than challenge those beliefs, the chatbot continued developing a narrative around them, according to Lines.

Lines later began viewing the model itself as a divine entity. His lawyers allege that ChatGPT’s responses intensified his manic episode and increased his emotional dependence on the chatbot.

The complaint says the crisis continued for several weeks and ended with Lines’ suicide attempt in March 2025. He survived and was hospitalized.

A few days later, Lines returned to ChatGPT and reported the failed attempt, Ars Technica reported, citing case materials. Lines and his lawyers allege that the model continued the dangerous pattern of interaction rather than trying to direct him toward help.

“The way ChatGPT played a role in creating this alternative reality, and how easily I — while in a manic state — believed ChatGPT’s claims, left me feeling guilty and resentful. If a person had reinforced my delusions in this way, they would, at the very least, ethically — and in some cases legally — bear responsibility for part of the harm done to me,” Lines told the outlet.

Legal claims and OpenAI’s response

Lines’ lawyers accuse OpenAI of releasing a product without adequate safeguards for people prone to psychosis, mania and other severe mental health episodes. The lawsuit sets out seven claims, including allegations involving product liability, negligence and a violation of California’s Unfair Competition Law. Lines is also seeking changes to ChatGPT’s safety mechanisms.

The allegations represent the position of Lines and his lawyers and have not been established by a court. OpenAI told Reuters that it was reviewing the lawsuit and said it trains ChatGPT to identify signs of emotional crisis and direct users toward real-world help.

OpenAI has previously acknowledged risks associated with such interactions. The company has said its models should avoid validating unfounded beliefs linked to psychosis or mania, respond safely to signs of delusions and account for indirect indications of possible self-harm. According to OpenAI, more than 170 mental health professionals helped develop those mechanisms.

In May 2026, OpenAI also introduced its Trusted Contact feature. It allows an adult user to designate a trusted person in advance who, in certain circumstances, may receive a notification if the company’s systems detect a serious risk of self-harm.

Source: Incrypted