LAS VEGAS (KLAS) — The Tesla Cybertruck bombing at the Trump International Hotel was the first time large language models have been used in an attack, according to Las Vegas Metropolitan Police Department (LVMPD) officials.
“We also have clear evidence in this case now that the suspect used ChatGPT AI to help plan this attack,” Sheriff Kevin McMahill said.
The tool used in the case of the New Year’s Day bombing is a large language model created by OpenAI called ChatGPT. The use of ChatGPT in planning the incident has raised questions about AI and how it manages harmful content.
Paul Keener, a Cyber Security Strategist with GuidePoint Security, said the technology is designed to be helpful.
“It’s designed to provide the best response that it can generate, not always factual but will provide a response that will be most logical,” Keener explained. “You can ask it how fast a 9mm bullet will go or other publicly available information, but if you ask it to create a diagram of a shape charge, it’s going to say no.”
Keener, for his part, believes the technology will improve with time.
“As we’re in the initial stage of AI, the regulations are a little looser, but I think the controls are going to get a lot tighter,” Keener said. “It’s going to get more factual in the way it responds and detects malintent better.”
OpenAI, the creator of ChatGPT, issued a statement, saying, “We are saddened by this incident and committed to seeing AI tools used responsibly. Our models are designed to refuse harmful instructions and minimize harmful content. In this case, ChatGPT responded with information already publicly available on the internet and provided warnings against harmful or illegal activities. We’re working with law enforcement to support their investigation.”
Keener posited that AI’s more significant threat lies in its ability to create narratives for scammers to use to steal money and data, advising individuals to avoid unsolicited emails and text messages and to have appropriate suspicion for anything that seems too good to be true.
LVMPD officials said they are sharing what they have learned from the bombing with other law enforcement agencies, detailing how large language models were used in the deadly incident.
