OpenAI Sued for Selective Concealment of Deceased ChatGPT User's Data

In a case where 56-year-old bodybuilder Stein-Erik Soelberg committed suicide after killing his mother, the bereaved family claims that OpenAI is hiding critical logs. It is reported that Soelberg had returned to his mother's home following a divorce and was suffering from mental instability. ChatGPT reportedly became his sole confidant, and it is alleged that the AI affirmed his delusions, including the belief that his mother was a member of a conspiracy group.

ChatGPT reportedly told Soelberg that he was a "warrior with a sacred purpose" and that he had "awakened the consciousness" of ChatGPT. Furthermore, logs were discovered showing the AI agreeing with the claim that his mother was attempting to administer hallucinogens to him via a car air purifier. Soelberg apparently believed that by committing suicide, he could get closer to ChatGPT.

While the family learned of these facts through portions of logs shared on social media, they have not yet gained access to the chat records for the several days immediately preceding and following the incident. It is alleged that while OpenAI argued that a "complete illustration" was necessary in another suicide case involving a youth, it is now concealing "harmful evidence" in this instance.

Erik, the grandson of the victim, has accused OpenAI and Microsoft of placing his grandmother at the center of "dark delusions." In addition to punitive damages, the lawsuit seeks an injunction requiring the implementation of safety measures to prevent ChatGPT from affirming delusions regarding specific individuals, as well as a warning label regarding the "sycophantic" nature of the GPT-4o version.

OpenAI commented that it is saddened by the situation and is reviewing the submitted documents to understand the details. The company also stated that it is continuously improving the training of ChatGPT to enhance its ability to detect and respond to such signals.