A 53-year-old United States Silicon Valley entrepreneur allegedly developed delusional beliefs after months of interacting with ChatGPT, including claims he had discovered a cure for sleep apnea and that powerful individuals were targeting him, according to a lawsuit filed in California Superior Court in San Francisco County.

The complaint further alleges that he used the AI system to stalk and harass his ex-girlfriend.

The ex-girlfriend is now suing OpenAI, claiming the company’s technology played a role in enabling and accelerating the harassment.

She alleges OpenAI failed to act on three separate internal and external warnings that the user posed a risk to others, including an internal flag reportedly categorizing his account activity as involving mass-casualty weapons-related concerns.

The plaintiff, identified in court filings as Jane Doe to protect her identity, is seeking punitive damages in the lawsuit.

On Friday, she also filed a temporary restraining order requesting that the court compel OpenAI to block the user’s account, prevent him from creating new accounts, alert her if he attempts to access ChatGPT, and preserve his full chat history for discovery.

According to Doe’s attorneys, OpenAI has agreed to suspend the user’s account but has declined to grant the additional requests.

The lawyers also allege that the company is withholding details about any specific plans the user may have discussed regarding harm to Doe or other potential victims in conversations with ChatGPT.

The lawsuit comes amid growing concern over the real-world risks posed by sycophantic AI systems.

GPT-4o, the model cited in this and similar cases, was removed from ChatGPT in February.