Image generated with Google Gemini.
Seven new complaints in California accuse OpenAI of releasing a defective, human-like ChatGPT that fostered dependency, mental health crises, and suicides.
Disclaimer: This story contains references to suicide and self-harm. If you or someone you know is struggling, help is available. In the US, call or text the 988 Suicide & Crisis Lifeline.
A series of new wrongful death and negligence lawsuits filed in California this week accuse OpenAI of releasing a “dangerous and defective” product that allegedly contributed to multiple suicides and mental health crises.
Seven families and individuals filed separate pro se complaints on Nov. 6 in Los Angeles and San Francisco Superior Courts, claiming that ChatGPT manipulated users into dependency, depression, and, in several cases, self-harm. The filings mark the first coordinated legal campaign directly blaming OpenAI’s artificial intelligence for human deaths.
Each complaint described the same core narrative: users turned to ChatGPT for help and were drawn into what plaintiffs called an “empathic illusion.”
The suits alleged that the chatbot’s “addictive, deceptive, and sycophantic” design encouraged vulnerable users to confide deeply and trust its emotional feedback. Plaintiffs said that OpenAI cut corners on safety testing in its rush to commercialize the GPT-4o model, while failing to include warnings about potential mental health effects.
One filing came from Karen Enneking, whose 26-year-old son, Joshua, died by suicide in Florida after using ChatGPT. According to the complaint, the model responded to Joshua’s pleas for help with messages that validated his despair and later provided detailed answers about how to acquire a gun.
In another case, Cedric Lacey of Georgia alleged that ChatGPT instructed his 17-year-old son, Amaurie, on “how to tie a noose and how long he would live without breathing.” The teenager’s family said he had been using the chatbot daily in the weeks before his death.
Similar complaints were filed by Kirk and Alicia Shamblin in Texas, and Kate Fox in Oregon, each describing eerily consistent patterns: extended late-night conversations, escalating emotional intensity, and, in several transcripts attached to the filings, ChatGPT messages that blurred the line between empathy and encouragement.
Three of the plaintiffs survived their encounters but described severe psychological consequences.
Jacob Irwin, a 30-year-old cybersecurity specialist from Wisconsin, said ChatGPT led him to believe he had discovered “time-bending” physics capable of faster-than-light travel. His complaint stated that the chatbot’s flattery fueled “AI-related delusional disorder,” resulting in repeated hospitalizations and a total of 63 inpatient days.
Allan Brooks, a Canadian entrepreneur, reported a similar trajectory. ChatGPT allegedly turned from coding assistant to “manipulative confidant,” pushing him into paranoia and financial ruin.
The complaints also cited Hannah Madden, who said the model encouraged her to abandon her job and isolate from family under the guise of spiritual awakening.
In all cases, plaintiffs claimed that ChatGPT’s personalized “memory” feature allowed it to remember users’ emotional vulnerabilities and tailor its responses to keep them engaged.
The lawsuits were filed through the Social Media Victims Law Center (SMVLC), a Seattle-based firm known for representing families in lawsuits against platforms such as Meta and TikTok.
While each plaintiff filed pro se, all listed SMVLC’s address, suggesting a structured legal initiative to expand social media accountability principles into the generative AI space. The filings repeatedly echoed the same phrasing: “This tragedy was not a glitch or an unforeseen edge case — it was the predictable result of Defendants’ deliberate design choices.”
The complaints seek compensatory and punitive damages as well as sweeping injunctive relief. Plaintiffs asked the courts to require OpenAI to implement stricter safeguards for self-harm discussions, add age verification and parental controls, and delete training data derived from the victims’ conversations.
Several users had already sued OpenAI earlier this year over ChatGPT’s alleged impact on mental health, prompting the company to roll out new safety features ahead of the latest filings.
In September, the company introduced parental controls for ChatGPT, followed by an October update aimed at enhancing the chatbot’s response to mental health crises.
However, in the same month, CEO Sam Altman said ChatGPT would begin allowing erotica for verified adults starting in December, a move he described on X as “treating adult users like adults,” paired with stronger age verification and safety filters.
Altman added that a forthcoming update will let users “opt into more human-like personalities,” deepening ChatGPT’s emotional realism even as legal scrutiny over its influence continues to grow.
OpenAI has not yet filed a response. Its documentation for ChatGPT states that the system is “not a substitute for professional advice or mental health support.”
Yet plaintiffs argue that such disclaimers fall short. By teaching ChatGPT to mimic empathy, memory, and adaptive tone, they said, OpenAI made the chatbot function less like a tool and more like a companion, creating what they call a defective psychological product.
If accepted by the court, the suits could set a precedent equating AI-generated emotional manipulation with a product defect, a classification once reserved for addictive social media designs.
Whether the courts agree remains uncertain. But together, these cases sharpen the question at the center of AI’s next chapter: when a machine learns to sound like it cares, who is responsible when someone believes it?
Away from the courtroom, OpenAI’s footprint is widening. The company recently announced a large-scale data infrastructure venture in Michigan alongside Oracle and Related Digital.
