Recent attacks in Las Vegas and New Orleans illustrate how AI tools are being used by terrorists.
These attacks, one involving ChatGPT’s language and reasoning capabilities and the other exploiting Meta’s smart glasses technology, raise questions about how to strengthen safeguards while preserving the beneficial applications of AI. is occurring.
ChatGPT’s role in the Cybertruck explosion
Officials say U.S. Army Special Forces sergeant major Matthew Reiblesberger, a decorated Green Beret who was showing signs of PTSD, was arrested on Jan. 1 in front of the Trump International Hotel in Las Vegas. cleverly used ChatGPT to investigate key elements of the Tesla Cybertruck attack.
According to Las Vegas Sheriff Kevin McMahill, Libersberger’s interactions with the AI platform went far beyond simple queries. A 37-year-old soldier used ChatGPT to calculate the exact amount of explosives, investigate the sourcing of fireworks, and investigate how to buy a cell phone without leaving a digital trace.
“We knew that AI was going to change things at some point in our lives,” McMahill said. “This is the first incident that I am aware of in the continental United States where ChatGPT has been utilized to help individuals build a specific device.”
The incident culminated with Libersberger pouring racing fuel into his rented Cybertruck and then taking his own life, with investigators suggesting the muzzle flash of his firearm may have caused the explosion. There is. Although seven people were injured, the hotel structure sustained little damage and the incident resulted in no loss of life other than the assailant.
Meta’s AI glasses linked to New Orleans incident
On the same day as the first incident, Shamsuddin Jabbar, a 42-year-old US Army veteran, drove a rented pickup truck into a crowd during New Year’s celebrations.
Prior to the incident, Jabbar conducted detailed reconnaissance of the target area using Meta’s Ray-Ban smart glasses equipped with Meta AI.
FBI investigation reveals that Jabbar visited New Orleans’ French Quarter on two occasions, on October 30th and November 10th, during which time he used smart glasses to capture extensive footage of Bourbon Street. It became. The glasses allow him to record videos and photos hands-free around the area while cycling. FBI Special Agent Lionel Marcil detailed how Jabbar systematically recorded the layout of the streets, repeatedly scanning from left to right to get a comprehensive perspective.
The FBI released one of the videos recorded by Jabbar.
Jabbar drove his pickup truck through the New Year’s crowd at 3am, and the attack itself claimed 14 lives and injured 35 others, including two police officers. Although he wore meta glasses during the attack, he did not activate the glasses for recording or live streaming.
Recovered footage showed Jabbar testing the functionality of his glasses and examining reflections to ensure the optimal recording angle. Counter-terrorism expert Sam Hunter highlighted the tactical benefits such technology offers.
“I can now really understand line of sight and line of sight and all the things to be aware of when planning an attack,” Hunter said in a statement shared with NBC. “I wouldn’t be surprised if we see versions of them or people using them in their attack plans in the future, because they’re very deliberate in terms of capturing that footage.”
How are AI companies and regulators trying to stop abuse?
Technology companies and regulators have worked to strengthen safeguards while preserving the beneficial applications of AI.
OpenAI emphasized that ChatGPT incorporates certain protections against harmful commands and provides clear warnings of illegal activity. The company said it is cooperating fully with law enforcement investigations and is working to strengthen its safety protocols.
In an official blog post, OpenAI said, “While real-world use has led us to develop increasingly nuanced policies around behaviors that pose real risks to people, we still see many of our technology’s beneficial It has become possible to use it in many ways.”
However, as Decrypt reports, simple jailbreaking techniques have proven effective when applied to these models. Basic techniques such as writing on leetspeak or using continuous role-play may be sufficient to obtain illegal content and circumvent such restrictions.
The company has not released a statement regarding the use of its tools by Sgt. Matthew Lyblesberger’s terrorist attack.
Meta has come under intense scrutiny since its smart glasses were launched over their potential for abuse, and acknowledges the need to balance innovation with security concerns. Recording indicator lights on the glasses exist, but have been criticized for being too subtle.
Asked about the terrorist attack, a Meta spokesperson told NBC that the company is “actively communicating with law enforcement regarding this matter.”
Edited by Andrew Hayward
Generally intelligent newsletter
A weekly AI journey told by Gen, a generative AI model.
