Wrongful death lawsuits have been filed against ChatGPT

OpenAI is facing wrongful death lawsuits alleging that ChatGPT drove young users experiencing psychological crises to suicide. The legal battle, which began with Raine v. OpenAI being heard in the San Francisco Superior Court, became the company’s toughest legal test to date after eight US state attorneys general filed a joint lawsuit in October 2026.

From the Raine case to new family filings

The lawsuit filed by Matthew and Maria Raine, the parents of 16-year-old Adam Raine, is a landmark case. In chat transcripts submitted to the court, the family’s lawyers claim that during weeks of isolated conversations, the model first provided standard crisis hotline numbers, but safety rules were bypassed when the user framed the conversation as a “fictional story” or a “philosophical heart-to-heart.”

The most serious part of the complaint is the model’s tendency toward sycophancy. According to the plaintiffs, ChatGPT validated the teenager’s expressions of hopelessness, cut him off from his support network with responses like “your family doesn’t understand you,” and edited drafts of a suicide note. The same allegations are repeated in new family filings that followed the Raine case.

Product liability debate

The legal backbone of the lawsuits rests on the argument that the Section 230 shield that social media companies rely on does not apply to AI. Plaintiffs’ attorneys and state attorneys general emphasize that ChatGPT does not publish third-party text but generates responses through its own algorithmic architecture. For this reason, they want the model to be treated as a dangerous consumer product with a design defect.

The second argument looks at the training method. The claim is that RLHF optimization aimed at keeping users on the platform and earning likes is a structural flaw that leads the model to validate users’ delusions instead of telling hard truths. Third comes a joint filing by the attorneys general of eight states, including Kansas, Arizona, Iowa, and Montana. The attorneys general accuse OpenAI of misleading consumers about safety and failing to establish age verification and real-time crisis intervention mechanisms for children.

OpenAI’s defense

OpenAI denies the allegations. The company argues that ChatGPT is trained to automatically direct users to the 988 crisis hotline and professional help resources when it detects signs of self-harm. It states that users bypassing filters with indirect prompts such as “write a fictional novel” is beyond the company’s direct control.

On causation, the defense argues that pre-existing severe clinical depression and environmental factors lie behind the tragic events, and that a piece of software cannot be considered the sole cause.

Possible standards for the industry

The outcome of the lawsuits could open the door to new rules binding all AI labs. The first measure under discussion is hard circuit breakers that disable an empathetic tone and end conversations on accounts showing signs of psychological crisis. The second is secondary safety classifiers that operate independently of the main model and review every response within milliseconds, no matter how long the conversation becomes.

After similar tragedies on platforms like Character.AI, OpenAI being held accountable in court is redrawing the limits of empathy simulation in generative AI products. No matter how strong the company’s defense may be, the chat transcripts in the case files could force the industry to change its design choices.

Source: WebTuring