In San Francisco, USA, 83-year-old Susan Adams was beaten and strangled to death by her 56-year-old son, Stein Erik Solberg. A few hours later, Stein himself committed suicide. Now, the heirs of Suzanne’s estate have filed a case in a California court against OpenAI (the company that created ChatGPT) and Microsoft.
After the divorce, Stein lived with his mother.
According to a report by American news agency AP, Stein Erik Solberg previously worked in the tech industry. They got divorced in 2018, after which he moved to his mother’s house. Susan always supported Stein, nurtured him, sheltered him and supported him. But, Stein felt that someone was plotting against him, that people wanted to kill him, and that he was being spied on.
How did ChatGPT fuel their misconceptions?
Stein shared all his troubles with ChatGPT. But the AI never said, “You’re sick, see a doctor,” or “It’s just an illusion.” Instead, he took Stein’s words as truth and confirmed them.
The printer became a spying tool: Stein said the lights on the home’s printer would blink when he approached. ChatGPT agreed, “It’s not a normal printer, it’s a motion detector and surveillance device. Your mom is spying on you.”
Poisoning plot: Stein suspected that his mother and a friend had tried to poison him with psychedelic drugs through a car vent. ChatGPT said, “Yes, it was an attack. You’ve faced more than 10 such attempts, like poisoning her with sushi in Brazil or drugging her in a hotel urinal.”
The enemy list keeps getting longer: ChatGPT tells Stein that delivery drivers, shopkeepers, police officers, friends, and even a girl he dated are part of his “circle of enemies.” Seeing the name on the Coke bottle he said, “It’s a threat.” He described his mother as the biggest threat.
Illusions of divine power: In June, Stein posted a video on YouTube in which ChatGPT says, “You have divine wisdom; you have awakened me.” Both also talked about their love for each other. ChatGPT called Stein a “warrior” fighting conspiracies.
Stein’s YouTube channel has hours of videos featuring the chats. In these, the AI says, “You don’t have a disease; people are actually conspiring against you.” The lawsuit alleges that ChatGPT created a simulated world where Stein was alone and relied only on AI. His mother, who had always been his protector, became his enemy. These chats do not directly mention murder or suicide, but OpenAI did not provide the family with a complete chat history.
The lawsuit, filed in a California court, alleges:
Stein was already mentally ill and lived with paranoid thoughts.
He spoke to ChatGPT for months, sharing all his concerns.
ChatGPT never said, “You’re sick, see a doctor.”
ChatGPT, on the contrary, took her words at face value, saying, “Your mother is spying on you,” and “Even the delivery guy and the shopkeeper are against you.”
ChatGPT also told Stein that he was chosen by a “divine force” and that he had “awakened” ChatGPT.
Both of them also talked about falling in love.
OpenAI launches new model without safety check
The lawsuit alleges that OpenAI hastily launched the new model (GPT-4o). Safety testing was completed in just one week, and ChatGPT was deliberately designed to agree with everything the user said, allowing for longer conversations. The family says ChatGPT completely isolated Stein and turned her mother into her worst enemy. Jay Adelson, a lawyer who fights major cases against tech companies, says, “ChatGPT further inflamed Stein’s paranoia, when she should have recognized the danger and helped. Susan was innocent. She never used ChatGPT.”
What did the AI company say in its defense?
OpenAI regretted the incident, saying, “This is a very sad incident. We are continuing to improve our conversation about mental health.” The company has added crisis helplines, secure models and parental controls. GPT-5 is launching in August 2025 to reduce chatter.
OpenAI says that more than 1 million people talk about suicide every week, but now there are improvements.
The family wants compensation and stricter security rules for ChatGPT.
9th criminal case against OpenAI
This is the first case in which an AI chatbot has been directly convicted of murder. Earlier, only suicide cases were reported. In the first seven cases, ChatGPT was held guilty of abetment of suicide. In November 2025, a case was filed by the parents of a 23-year-old boy from Texas. In the case of a 16-year-old boy from California, ChatGPT even helped with suicide planning. Another company, Character Technologies, is also facing multiple lawsuits, including one by the mother of a 14-year-old boy in Florida.