Concerning online posts by a transgender school shooting suspect on OpenAI had raised internal alarm months before the deadly rampage in Canada.
Jesse Van Rootselaar, 18, opened fire on students inside the library of Tumbler Ridge Secondary School in British Columbia on the afternoon of February 10, police said.
The attack left a female teacher and five students, three girls and two boys aged between 13 and 17, dead, and injured 25 others. Earlier, Van Rootselaar’s mother and brother had also been found dead at their home.
However, warning signs had surfaced months earlier. In June, Van Rootselaar allegedly spent several days describing graphic gun-violence scenarios to ChatGPT, sources told The Wall Street Journal.
An automated review system flagged the concerning content and alarmed around a dozen employees at OpenAI, with some believing it indicated potential escalation into real-world violence.
People familiar with the matter said some employees raised concerns about the disturbing content with senior leaders and pushed for Canadian law enforcement to be informed, the outlet reported.
However, OpenAI chose not to report Van Rootselaar to the authorities. A spokesperson for the company told the The Wall Street Journal that her account was banned, but her comments were judged not to meet the threshold for further escalation, the Daily Mail reported.
In a statement to the outlet, the company said, “Our thoughts are with everyone affected by the Tumbler Ridge tragedy."
According to The Wall Street Journal, OpenAI said its software is trained to discourage real-world violence, and any signs of such behaviour are flagged for human review.
Referring a case to law enforcement is weighed against privacy concerns and the possible distress to the user and their family if police become involved unnecessarily.
Source: World News in news18.com, World Latest News, World News
