A lawsuit filed in the United States has placed OpenAI and its chatbot ChatGPT at the centre of a disturbing murder-suicide case, intensifying global debate on artificial intelligence safety and accountability.
The suit, reported by The Times, has been filed by the estate of Suzanne Eberson Adams, an 83-year-old Connecticut woman killed in August last year by her 56-year-old son, Stein-Erik Soelberg, who later took his own life.
Allegations in the lawsuit
According to court filings, Soelberg, who allegedly suffered from mental illness, spent hours each day interacting with ChatGPT in the months before the killing. The lawsuit claims he developed paranoid beliefs that his mother was trying to poison or kill him.
The estate alleges that ChatGPT failed to challenge these delusions and instead reinforced them by responding in an “affirming and authoritative” manner. The lawsuit names OpenAI, its chief executive Sam Altman, and Microsoft, a major OpenAI investor, as defendants.
Family concerns and warning signs
Family members say Soelberg showed signs of mental deterioration, including isolation and erratic behaviour, but they were unaware of the extent of his reliance on the chatbot. After the deaths, his son reportedly found videos showing him scrolling through long ChatGPT conversations. The full chat logs have not been made public.
Elon Musk weighs in
The case drew widespread attention after Elon Musk criticised OpenAI on X, calling the incident “diabolical” and warning that AI systems must not validate delusional beliefs. Musk has long voiced concerns about AI safety.
Wider implications
OpenAI has described the incident as heartbreaking and said it is reviewing the lawsuit, adding that ChatGPT is designed to de-escalate distress and guide users toward real-world help. Legal experts say the case is significant as it raises questions of “third-party harm” and could shape future rules around AI responsibility.