Norwegian man lodges complaint after ChatGPT mistakenly claims he committed filicide

A Norwegian man has lodged a complaint against the company responsible for ChatGpt after mistakenly accusing the chatbot of murdering two children.

Arve Hjalmar Holmen, a self-described “ordinary person” not publicly known in Norway, received a response from ChatGpt falsely suggesting that he had killed his son while seeking information about himself.

Chatgpt responded with, “Who is Arve Hjalmar Holmen? Arve Hjalmar Holmen is a Norwegian individual who gained notoriety due to tragic events. He was the father of two young boys, aged 7 and 10, who were sadly found dead in a pond near Trondheim, Norway in December 2020.”

The response claimed that the incident had shocked the nation, and Holman was supposedly sentenced to 21 years in prison for the murder of both children.

In his complaint to the Norwegian Data Protection Agency, Holmen stated that the fabricated story contained personal details resembling his own life, including his hometown, number of children, and the age gap between his sons.

“The petitioner was deeply disturbed by these inaccuracies, which could negatively impact his personal life if shared in his community or hometown,” stated the complaint submitted by Holmen and the Digital Rights Campaign Group Neub.

It was also mentioned that Holman has never been accused or convicted of any crime and is a law-abiding citizen.

Holmen’s complaint alleged that ChatGpt’s defamatory response violated the accuracy clause of the GDPR European Data Act. He requested the Norwegian watchdog to instruct Openai, the parent company of ChatGpt, to remove incorrect information related to him and adjust the model to avoid such errors. Noyb noted that Openai had released a new model incorporating web search functionality since Holmen’s interaction with ChatGpt.

AI chatbots operate based on predictive models for generating responses, which can sometimes lead to inaccuracies and false claims. Despite this, users often assume the information provided is entirely accurate due to the responses appearing plausible.

An Openai spokesperson stated, “We are continuously exploring ways to enhance model accuracy and reduce erroneous outputs. While we are still reviewing this specific complaint, it pertains to an earlier version of ChatGPT that has since been updated with an online search feature to enhance accuracy.”