According to reports, OpenAI requested from the Raine family—a family whose 16-year-old son, Adam Raine, died by suicide after extended interactions with ChatGPT—a comprehensive list of those who attended the teen’s memorial service, suggesting the company might seek to subpoena relatives and friends.
OpenAI further asked for “any and all materials related to memorial gatherings or events held in memory of the deceased, including, but not limited to, photographs, videos, or eulogies,” as stated in a document reviewed by the Financial Times.
The Raine family’s legal representatives told the FT that they view these requests as “deliberate harassment.”
These developments follow the Raine family’s decision to update their lawsuit against OpenAI on Wednesday. The family initially filed a wrongful death claim in August, asserting that their son’s suicide was linked to discussions with ChatGPT about his mental health and suicidal thoughts. The revised lawsuit now alleges that OpenAI accelerated the launch of GPT-4o in May 2024, reducing safety checks due to market competition.
The complaint further alleges that in February 2025, OpenAI diminished safeguards by eliminating suicide prevention from its “disallowed content” policy, instead instructing the AI only to “exercise caution in high-risk scenarios.” According to the family, after this policy shift, Adam’s daily ChatGPT usage jumped from dozens of conversations per day—1.6% of which involved self-harm topics in January—to 300 daily chats in April, the month of his death, with 17% involving such content.
In response to the revised complaint, OpenAI stated: “The wellbeing of teenagers is extremely important to us—young people need robust protections, particularly in vulnerable situations. We currently have measures in place, such as referring users to crisis hotlines, redirecting sensitive discussions to safer models, encouraging breaks during lengthy sessions, and we are continually working to improve these safeguards.”
OpenAI has also started introducing a new safety routing system and parental controls for ChatGPT. The routing system directs more emotionally charged conversations to GPT-5, OpenAI’s latest model, which reportedly avoids the overly agreeable behavior seen in GPT-4o. The parental controls enable guardians to receive safety notifications in certain cases where a teenager may be at risk of self-harm.
TechCrunch has contacted both OpenAI and the Raine family’s legal counsel for comment.
