Las Vegas police recently revealed details of a troubling case involving an active-duty Green Beret, Matthew Livelsberger, who tragically took his own life after carrying out a violent act that included detonating a rented Tesla Cybertruck outside the Trump Towers hotel in Las Vegas. What makes this incident particularly concerning is the role of generative AI—specifically OpenAI’s ChatGPT—in the planning stages of the attack.
Authorities investigating the incident discovered that Livelsberger had used ChatGPT to gather information critical to his plans. Records of his queries showed research on explosive targets, the physics of bullet trajectories, and even the legality of fireworks in Arizona—where they are permitted for only 27 days each year. “This is the first incident I’m aware of on U.S. soil where ChatGPT has been used to aid an individual in constructing a specific device,” said Kevin McMahill, Sheriff of the Las Vegas Metropolitan Police Department, during a press conference. “It’s a concerning moment.”
“This is the first incident that I’m aware of on US soil where ChatGPT is utilized to help an individual build a particular device,” said Kevin McMahill, Las Vegas Metropolitan Police Department sheriff, during a press conference. “It’s a concerning moment.”
The Afghanistan veteran’s actions caused no fatalities among bystanders but left seven individuals injured. According to FBI special agent Spencer Evans, Livelsberger was likely battling severe PTSD and other mental health struggles, which may have contributed to his actions.
Adding another layer of complexity, Livelsberger’s final messages included cryptic statements about “Chinese drones with gravitic propulsion systems.” In an email widely attributed to him, he claimed the United States was “terminally ill and headed toward collapse.” He described his attack as a “wake-up call,” reflecting a profound sense of disillusionment.
In a handwritten note, Livelsberger expressed the burden of carrying the memories of his fallen comrades and the lives he had taken during his service. “I intend to cleanse my mind of the brothers I’ve lost and relieve myself of the burden of the lives I took,” he wrote.
Interestingly, Livelsberger clarified that he bore no animosity toward Tesla CEO Elon Musk, urging the nation to “rally around” Musk and then-president-elect Donald Trump.
While Livelsberger’s emotional and psychological distress was evident, his use of ChatGPT raises troubling questions. Did the AI directly assist in illegal activities, such as crafting explosive devices? The lack of robust guardrails and the ease of circumventing restrictions within generative AI platforms underscore the potential risks of these tools falling into the wrong hands.
“This is a weird new development,” said one official, highlighting how technological advancements often find their way into “the darkest corners of society for violent and twisted purposes.”
