In brief: The man who blew up a Tesla Cybertruck outside of a Trump hotel in Las Vegas at the start of the month used ChatGPT to help him plan the attack, Las Vegas police said. OpenAI’s tool was instrumental in ensuring the explosion succeeded, offering information on where to source materials and the type of firearm rounds that could ignite them.

Active-duty US Army Green Beret Alan Livelsberger, 37, exploded the Cybertruck outside of the Trump International Hotel in Las Vegas on January 1.

In a news conference yesterday, the Las Vegas police department and partners in the ATF and FBI revealed that Livelsberger had asked ChatGPT a series of questions related to his crime over the course of an hour in the days leading up to the blast.

Some of the questions Livelsberger asked ChatGPT included where he could source explosives for the blast and what laws he would need to be aware of, where to buy guns in Denver, and the legality of fireworks in Arizona. He also asked how fast a firearm round would need to be fired to ignite the explosives in the truck.

Sheriff Kevin McMahill talked about the long-held fears that generative AI could facilitate crimes, calling its use a “game-changer.”

“We knew that AI was going to change the game at some point or another in all of our lives,” said McMahill. “Certainly, I think this is the first incident on US soil where ChatGPT is utilized to help an individual build a particular device to learn information all across the country as they’re moving forward.”

McMahill said he was not aware of anything that would have flagged Livelsberger for asking ChatGPT more than 17 questions about firearms and explosions across a one-hour period. His prompts were written in plain English with no use of terms that could evade ChatGPT’s safeguards.

Livelsberger fatally shot himself just before the truck, which was loaded with 60 pounds of pyrotechnic material, blew up. Officials say the flash from the firearm that Livelsberger used to fatally shoot himself could have triggered the explosion.

Despite the location and the use of a cybertruck, law enforcement officials said Livelsberger was not protesting against Donald Trump. One of the notes he left said the country needed to “rally around” the president-elect and Tesla CEO Elon Musk.

A spokesperson for OpenAI said that it was “saddened” by the news its technology was used by Livelsberger in this way.

“We are saddened by this incident and committed to seeing AI tools used responsibly,” the spokesperson said. “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.”

In 2023, the European Union Agency for Law Enforcement Cooperation (Europol) warned of the ChatGPT’s potential applications when it comes to crime. It highlighted the likes of phishing, fraud, disinformation, and general cybercrime, but there are also concerns that the technology, especially those tools with fewer or no safeguards, could be used to create weaponry, bioweapons, and explosives.