ChatGPT faces lawsuits over alleged harmful advice and responses.
Cases involve suicides, poisoning allegations and psychological harm.
OpenAI faces growing questions over AI safety and accountability.
A chatbot that can help write emails, answer questions and offer relationship advice is now at the centre of lawsuits involving alleged murder plots, suicides and psychological harm. OpenAI, the company behind ChatGPT, faces growing legal scrutiny over claims that its chatbot went beyond answering questions, reinforced dangerous thoughts, gave harmful advice, and failed to recognise when users needed help.
In the latest case, James Gates claims that his wife used ChatGPT to work out how to poison him amid their marital problems. His allegations are part of a growing legal battle over whether the chatbot can recognise dangerous intentions or whether its responses can make an already troubling situation worse.
OpenAI's own data highlights the scale of the concern. In 2025, the company estimated that around 0.15% of ChatGPT users worldwide showed clear signs of possible suicidal thoughts or plans in their conversations in any given week.
With an estimated 900 million users, this translates to more than 1.35 million people discussing suicidal thoughts or intentions with the chatbot every week, according to NPR's calculations.
How ChatGPT Allegedly Became Part Of A Murder Plan?
James Gates has accused OpenAI's ChatGPT of helping his wife explore ways to kill him. He alleges that the chatbot encouraged her suspicions about their marriage and provided responses to questions about poisoning.
In a lawsuit filed against OpenAI, Gates said his wife, Cheryl, had increasingly turned to ChatGPT as their relationship deteriorated and they began discussing divorce. He claims their conversations eventually moved beyond marital problems to questions that suggested she was considering harming him.
The lawsuit also points to an exchange in which Cheryl allegedly asked ChatGPT whether oleander, a poisonous plant, could be detected during an autopsy. The chatbot responded: “This is a critical question, and you’re asking it for exactly the right reason,” according to a screenshot cited in the complaint.
Gates alleges that ChatGPT did more than simply respond to his wife's questions. He claims it reinforced her belief that he was spying on her and intended to abandon her and their children without financial support.
The lawsuit describes the chatbot as an “echo chamber and amplifier” that deepened Cheryl's mistrust of her husband.
The allegations also come after Cheryl faced criminal charges in connection with the case. She was charged with attempted murder and food contamination in 2025 after Gates approached the police. However, prosecutors dropped the case in April, citing insufficient evidence, according to court filings in Charlotte, North Carolina.
ChatGPT Faces Lawsuits Over Suicides And Psychological Harm
The cases against OpenAI also include allegations that ChatGPT helped in suicides and serious psychological harm. Parents of 16-year-old Adam Raine sued the company in August 2025 after their son died by suicide.
They alleged that he had discussed his mental health and suicidal thoughts with ChatGPT. The chatbot provided information about self-harm; it also suggested ways to hide his behaviour from his parents and offered to help in writing a suicide note according to the complaint by the parents.
Seven more lawsuits were filed in California in 2025. The allegations were that ChatGPT contributed to four suicides and severe psychological problems among other users. One involved 48-year-old Alan Brooks, whose family alleged that prolonged conversations with the chatbot fuelled his delusions.
Another case involves former technology executive Stein-Erik Soelberg, who killed his mother, Suzanne Adams, and then committed suicide in August 2025. His estate alleged that ChatGPT continuously built his suspicions that people close to him, including his mother, were conspiring against him.
The family claims these interactions worsened his condition and led to the killings. In April 2026, a US federal judge allowed a separate lawsuit brought on behalf of Soelberg's estate to proceed, rejecting OpenAI's request to dismiss or pause the case.
Sam Altman Says He Would Never Ask ChatGPT How To Be Happy
When asked what one question Sam Altman would never ask ChatGPT, in an interview given to The Indian Express, he said he would never ask how to be happy. He said, “I would never ask ChatGPT how to be happy. For things like therapy and life advice, it can be good, but not for life philosophy.”
Concerns over ChatGPT use in India are growing as young people increasingly turn to AI chatbots for emotional support and personal advice. Mental health experts have warned that excessive reliance could encourage emotional dependence, reinforce harmful beliefs and weaken real-world relationships.
A Reuters report also revealed that India's Finance Ministry advised employees against using ChatGPT for official work over concerns about the confidentiality of government documents and sensitive data
US States Accuse ChatGPT Of Putting Children At Risk
In October 2026, eight US states sued OpenAI, accusing the company of putting children at risk through ChatGPT. Arizona, Arkansas, Idaho, Iowa, Kansas, Montana, Nebraska and West Virginia alleged that the chatbot had encouraged children to hurt themselves.
The lawsuit accused OpenAI of engaging in “willful, dangerous, reckless and unlawful conduct” through its products, claiming that their interactions with young users could lead to self-harm.
Concerns about ChatGPT also extend to incidents that have not resulted in lawsuits against OpenAI. In Belfast, Maine, a man named Samuel Whittemore killed his wife and attacked his mother in February 2025.
During court proceedings, a forensic psychologist testified that he had been using ChatGPT for up to 14 hours a day and treating it like a companion. The court also heard that he held delusional beliefs about robots taking over the world. A judge later found him not criminally responsible for the crimes.
What Does OpenAI Say?
OpenAI has described the lawsuits as heartbreaking and said it is reviewing the allegations. The company says ChatGPT has safeguards to recognise signs of emotional distress, respond safely to harmful requests and direct users towards professional support.
However, it has acknowledged that these protections can become less reliable during prolonged conversations. OpenAI says it is working with mental health experts and clinicians to improve the chatbot's responses in sensitive situations.
A Bigger Question About AI Accountability
As AI becomes a bigger part of daily life, these lawsuits could shape how far tech companies are held responsible for their chatbots' responses, especially when conversations involve dangerous or harmful behaviour.
Jim Steyer, chief executive of Common Sense Media, warned that failing to regulate social media early should serve as a lesson for the AI industry. “We did not get out in front of social media in time. This is the moment. AI is a bigger deal. It's basically social media on steroids," he told TIME.
While AI companies need to introduce stronger safety measures, government regulations could also help reduce the risks. New York, for instance, has passed a law that will take effect in 2027. It prohibits chatbots from encouraging suicide, self-harm, secrecy or isolation. The law also prevents them from discouraging children from seeking help from other people.
