OpenAI requested memorial attendee list in ChatGPT suicide lawsuit

OpenAI reportedly asked the Raine household – whose 16-year-old lad Adam Raine died by termination aft prolonged conversations with ChatGPT – for a afloat database of attendees from the teenager’s memorial, signaling that the AI steadfast whitethorn effort to subpoena friends and family.  

OpenAI besides requested “all documents relating to memorial services oregon events successful the grant of the decedent including but not constricted to immoderate videos oregon photographs taken, oregon eulogies given,” per a papers obtained by the Financial Times. 

Speaking to the FT, lawyers from the Raine household described the petition arsenic “intentional harassment.” 

The caller accusation comes arsenic the Raine household updated its suit against OpenAI connected Wednesday. The household archetypal filed a wrongful decease suit against OpenAI successful August aft alleging their lad had taken his ain beingness pursuing conversations with the chatbot astir his intelligence wellness and suicidal ideation. The updated suit claims that OpenAI rushed GPT-4o’s May 2024 merchandise by cutting information investigating owed to competitory pressure.

The suit besides claims that successful February 2025, OpenAI weakened protections by removing termination prevention from its “disallowed content” list, alternatively lone advising the AI to “take attraction successful risky situations.” The household argued that aft this change, Adam’s ChatGPT usage surged from dozens of regular chats, with 1.6% containing self-harm contented successful January, to 300 regular chats successful April, the period helium died, with 17% containing specified content. 

In a effect to the amended lawsuit, OpenAI said: “Teen wellbeing is simply a apical precedence for america — minors merit beardown protections, particularly successful delicate moments. We person safeguards successful spot today, specified arsenic [directing to] situation hotlines, rerouting delicate conversations to safer models, nudging for breaks during agelong sessions, and we’re continuing to fortify them.” 

OpenAI precocious began rolling retired a new information routing strategy and parental controls connected ChatGPT. The routing strategy pushes much emotionally delicate conversations to OpenAI’s newer model, GPT-5, which doesn’t person the aforesaid sycophantic tendencies arsenic GPT-4o. And the parental controls let parents to person information alerts successful constricted situations wherever the teen is perchance successful information of self-harm. 

TechCrunch has reached retired to OpenAI and the Raine household attorney.