Interactions With Chatbot Convinced Greenwich Man To Commit Murder: Reports

GREENWICH, CT — The artificial intelligence chatbot ChatGPT appears to have fed the delusions of a 56-year-old Greenwich man, who killed his 83-year-old mother and then himself, in a murder-suicide, according to several high-profile published reports.

According to reports in The Wall Street Journal, New York Post and Greenwich Time, Stein-Erik Soelberg, who at one time was a manager at Yahoo, considered ChatGPT to be his best friend, which Soelberg referred to as "Bobby."

Soelberg interacted with the chatbot for months, posting videos about the interactions on YouTube, before he killed his mother, Suzanne Adams, and himself.

The bodies of Soelberg and Adams were discovered on Aug. 5 in the Old Greenwich home they shared when Greenwich police conducted a welfare check on them.

Soelberg appeared to have been in a years-long battle with mental illness, which reportedly included his ex-wife obtaining a restraining order against him, and Soelberg also having multiple brushes with the law, according to the New York Post.

The case is reportedly one of the first of its kind, where ChatGPT may have been involved in allegedly egging on a disturbed individual to such a fatal conclusion, The Wall Street Journal reported.

Companies that are developing artificial Intelligence are grappling with how to make chatbots more human, but also keep them safe, according to the WSJ.

Greenwich Time reports that Soelberg posted two-dozen videos on his YouTube channel, on which he described himself in grandiose terms, according to Greenwich Time.

"Shaman Warrior of The Light," Soelberg wrote in the description on the YouTube channel. "Matrix eraser. Chosen by God’s grace to be heal/learn/share & teach. Blessed to share my path of redemption. Lead with Love."

The chatbot, developed by OpenAI, reportedly fed Soelberg's psychosis, telling him that he was a threat to the Matrix, and that a computer printer in the home may have also been spying on him, the Post reported.

Soelberg's interactions with the chatbot took on a more personal tone, in part because he reportedly used a feature that allowed it to remember previous chats between the two, according to the WSJ.

Additionally, the chatbot told him to look for symbols in a receipt from a Chinese restaurant linking his mother to a demon, Greenwich Time reported.

In a statement to the WSJ, OpenAI said the company has been in contact with Greenwich police about the case, and the company said in a recent blog post that it will introduce features to help people who may be experiencing a mental crisis.