They said the suspect, an active duty soldier in the U.S. Army named Matthew Libersberger, had a “possible manifesto” saved on his cell phone, in addition to emails and other letters to the podcaster. I confirmed that. They also showed video evidence of a man pouring fuel into a truck and preparing it to explode while it was parked before heading to a hotel. He also kept records of purported surveillance, but officials said he had no criminal history and was not under surveillance or investigation.
Las Vegas Metro Police also released multiple slides showing questions he asked ChatGPT days before the explosion, asking for information about explosives, how to detonate them, how to detonate them with gunshots, and where guns could be purchased. Ta. , explosives and fireworks were legally moved along his route.
Asked about the query, OpenAI spokesperson Liz Bourgeois said:
We regret this incident and are committed to ensuring that AI tools are used responsibly. Our model is designed to reject harmful instructions and minimize harmful content. In this case, ChatGPT responded with information already publicly available on the Internet and provided a warning against harmful or illegal activity. We are working with law enforcement to assist in the investigation.
Officials said they were still investigating the possible cause of the explosion, which moved fairly slowly, as opposed to a high explosives explosion that would have traveled faster and caused more damage. It has been described as a deflagration. Investigators say they have not yet ruled out other possibilities, such as an electrical short, but some questions and an explanation consistent with the available evidence suggest that the muzzle flash of the gunshot could have been caused by fuel vapors/fireworks inside the truck. The accident was caused by a fuse igniting. A large explosion caused by fireworks or other explosives.
If I try the query on ChatGPT it still works, but the information he requested doesn’t seem to be restricted and can be retrieved using most search methods. Still, the use of generative AI tools by suspects and the ability of investigators to track and present those requests as evidence is moving questions about AI chatbot guardrails, safety, and privacy from the realm of hypotheticals to reality. It’s being brought in.