Gephardt wonders if you can protect yourself from scams and frauds with AI

KSL TV

Save Story

Estimated read time: 4-5 minutes

This archived news story is available only for your personal, non-commercial use. Information in the story may be outdated or superseded by additional information. Reading or replaying the story in its archived form does not constitute a republication of the story.

SALT LAKE CITY — When it comes to online scams, some seem obvious. No, you didn't win the lottery. No, a Nigerian prince isn't asking you to help him get his fortune out of his country. And no, a famous Hollywood celebrity isn't romancing you while also hitting you up for money.

But other scams are "more sophisticated" to say the least. And in artificial intelligence, scammers have found an effective new tool.

All the time, I report on people duped into thinking they were talking to their bank, a well-known and reputable company, or even a loved one in trouble. Data from the Federal Trade Commission shows these sophisticated cons are costing Americans more and more every year.

But if the bad guys can use AI against us, can we use it to fight back?

Earl Foote says the AI chatbot did a “bang-up job” of alerting users to potential scams in a general sense but lacks the dialogue needed to help in specific situations.
Earl Foote says the AI chatbot did a “bang-up job” of alerting users to potential scams in a general sense but lacks the dialogue needed to help in specific situations. (Photo: Aubrey Shafer, KSL-TV)

For the answer, I went to Nexus IT Consultants where owner Earl Foote makes his living by helping clients keep scammers and hackers at bay. He agreed to help me test one popular AI program, ChatGPT. And we started by feeding it an easy, straightforward scenario.

"Say, I just got a package," I hypothesized with Foote. "I just got a text message saying I've got a package, but they can't deliver it and they're asking for personal information."

Foote read back ChatGPT's response to our inquiry.

"No, you're not to provide your personal information in response to that text message," he read. "This is likely a form of smishing a type of fishing."

The chatbot went on to describe how the con works and how we could respond.

"I think that's a really good answer," Foote said of the information from ChatGPT.

Our next scenario got a little trickier with a situation that often turns out to be a scam but also could be real. Say my friend and KSL colleague Sloan Schrage lost his shirt on the slot machines.

"Sloan could be calling you and saying, 'Hey Matt, I got stranded in Wendover. I'm gonna need 150 bucks to get back,'" Foote theorized.

He says this particular type of scam, often referred to as the "grandparent scam," is growing in popularity with the advent of AI mimicking voices. So, here's what ChatGPT thought of this scenario:

"It's important to verify the situation before sending any money," Foote read from the results.

Again, ChatGPT gave a long thread of information and things I could do to verify that it really is my buddy calling me for help, and not some fraudster.

"So, the same sort of things I would tell," Foote said of these results.

Ransomware

Let's try something a little different — ransomware. That's when a thief hacks into a device and locks the rightful owner out of it. For you and me, it means losing access to our precious photos or important files. For companies, it means losing millions of dollars.

"My computer is locked, and they say I have to send Bitcoin to unlock it. What should I do?" we asked ChatGPT.

Earl Foote warns the growth and accessibility of AI is giving cybercriminals a leg up in crafting harder to detect scams.
Earl Foote warns the growth and accessibility of AI is giving cybercriminals a leg up in crafting harder to detect scams. (Photo: KSL-TV)

The bot's advice? Do not pay the ransom.

"That is absolutely what the FBI recommends," Foote said of the results.

"But I personally know people who have lost things by not paying the ransom, and I know others who paid the ransom and got some access back," I shot back.

Foote says that for people who pay, the success rate is about 75%. And this also demonstrates one place where AI can fall short — there really isn't a dialogue.

"But if I call you and say, 'Earl, I'm desperate. … I've got to have this file,'" I asked. "What do you say to me?"

"We'll assess it with you again and the way we would approach it is what is the data worth to you," he answered.

ChatGPT's suggestion of calling a professional, however, was way down on its list of steps. Foote says that would be his first, immediate step.

"We do have some tools particularly for the most well-known ransomwares," he said. "We're able to decrypt those encrypted files sometimes without you paying a ransom, right?"

In the end, though, Earl thinks that ChatGPT proved a handy tool we can use to at least point us in the right direction when we engage with a potential scammer.

"For cybersecurity advice," he said, "it did a bang-up job."

Most recent Science stories

Related topics

KSL.com Beyond Series
KSL.com Beyond Business

KSL Weather Forecast

KSL Weather Forecast
Play button