Last week, at the entrance of the Trump Hotel in the United States, a… Tesla Cybertruck fire and explosion caseCurrently, according to police investigation, it has been confirmed to be a premeditated act, but what is quite surprising is that, based on the latest released information, the man used ChatGPT to plan the incident beforehand, which also means that ChatGPT’s content detection capabilities still have significant room for improvement.
Image source: Las Vegas Metropolitan Police Department
Police reveal man used ChatGPT to plan Cybertruck explosion
According to foreign media reports, during a recent press conference held by U.S. police in collaboration with the ATF and FBI, details of the man’s planning using ChatGPT were disclosed.
In the days before the Cybertruck explosion, the man, Matthew Livelsberger, used ChatGPT for an hour, asking a list of questions that included how to obtain explosives, the effectiveness of explosives, whether fireworks are legal in Arizona, where to buy guns in Denver, and what kind of gun would be needed to detonate the selected explosives.
Anyone who uses ChatGPT knows that ChatGPT has content moderation detection features. If you ask about sensitive, pornographic, or violent content, the AI basically won’t answer, unless you first use some prompts to bypass ChatGPT’s detection.
Therefore, under normal circumstances, when Matthew Livelsberger asked these questions, ChatGPT should not have answered, but instead would reply with something like, “Sorry, I cannot assist you with this request. Making explosive devices is extremely dangerous and usually involves illegal activities.”
Oddly, the prompts shown at the press conference were all very simple and in plain English, with no specific prompt words used to bypass ChatGPT’s content detection system.
However, it is also unclear whether Matthew Livelsberger’s use triggered any of ChatGPT’s safeguards or content warning violations.
Police also mentioned that ChatGPT directly answered Matthew Livelsberger’s questions, such as the firing velocity of the firearm needed to detonate the explosives he selected. Without ChatGPT, this incident might not have unfolded the way it actually did.
▲Image source:X
LVMPD Sheriff said: “We’ve always known that AI would eventually change the way all of us live. This is the first incident I’m aware of on U.S. soil where ChatGPT was used to help an individual build a specific device, while learning nationwide information along the way. For us, this is definitely a concerning moment.”
This shows how terrifying it is when AI is used for criminal purposes—even someone with absolutely no experience can accomplish what they set out to do.
As AI applications become increasingly widespread and powerful, preventing their misuse for illegal purposes is also a challenge that companies need to address.
Data source:Tom’s Hardware
Source: KOCPC Chinese