Ahead of the Democratic Party's New Hampshire primary on January 23, potential voters began receiving phone calls with a fake voiceover of President Biden urging them not to vote until the November general election. Slovakian voter audio posted on Facebook included fake AI-generated audio of a presidential candidate trying to steal the election, which could have resulted in an election that favored another candidate. Recent elections in Indonesia and Taiwan were also marred by AI-generated misinformation.
In response to the fake Biden robocalls in New Hampshire, the Federal Communications Commission decided on February 8 to outlaw AI-generated voices in robocalls. But experts IEEE spectrum Even though generative AI puts a new spin on age-old robocall scams and presents an opportunity to step up efforts to scam individuals, there's no confidence that the move will go far enough.
Total losses due to fraud and spam in the United States by 2022 are estimated to be US$39.5 billion, according to TrueCaller, which makes caller ID and spam blocking apps. According to research by Hiya, a company that provides currency protection and identity verification services, the average amount lost by people scammed in the U.S. that same year was $431.26. Hiya says that volume will increase as the use of generative AI gains momentum.
“Collectively, it’s amazing how much money is lost due to scams perpetuated through robocalls,” said Eric Burger, director of research for Virginia Tech’s Commonwealth Cyber Initiative.
“I don’t think we can assess how quickly the phone experience will change because of this.” —Jonathan Nelson, Hiya
AI makes it easier for fraudsters to target individuals.
“The biggest fear about generative AI is that it will make personalized fraud mainstream,” says Jonathan Nelson, director of product management at Hiya. In particular, he said generative AI would make spear phishing attacks easier to carry out.
Cost of Telephone Fraud
Average amount lost by phone scam victims in 2022 (in USD):
- USA: $431.26
- UK: $324.04
- Canada: $472.87
- France: $360.62
- Germany: $325.87
- Spain: $282.35
Source: Hiya
Phishing attacks typically aim to trick people into revealing personal information, such as passwords or financial information. However, spear phishing is more targeted. Scammers know exactly who they are targeting and hope to reap greater rewards through a more tailored approach. Now, with generative AI, fraudsters can even scrape social media sites, draft text, and clone trusted voices to rip off unsuspecting individuals out of their money en masse, Nelson says.
With the FCC's unanimous vote to outlaw AI-generated robocalls, the issue naturally shifted to enforcement. This is what the experts say IEEE spectrum Although many see this as a necessary first step, it is generally viewed with suspicion. “This is a helpful step, but it’s not a complete solution,” said Daniel Weiner, director of the Brennan Center’s Elections and Government Program. Given the scope of the FCC's authority, Weiner said it would be difficult for the agency to take a broader regulatory approach in the same vein as the general ban on deepfakes that the European Union is considering.
Burger, who was the FCC's chief technology officer from 2017 to 2019, said the agency's vote will ultimately only have an impact if it begins enforcing a ban on robocalls more generally. Most types of robocalls have been banned since the agency enacted the Telephone Consumer Protection Act in 1991. (There are some exceptions, such as prerecorded messages in a dentist's office, such as reminders of upcoming appointments.)
“I don’t think there’s any enforcement,” Berger said. “Politicians want to say, ‘We’re going after the bad guys,’ but that’s not the case. There’s no vibrancy that we want to see.”
Robocall enforcement tools may not be sufficient for AI.
A key way to identify the source of a robocall and prevent malicious actors from continuing to make calls is to trace the call through a complex network of communications infrastructure and identify where the call originates. Since calls typically pass through infrastructure managed by multiple network operators, such as AT&T and T-Mobile, backtracking has been a complex issue. However, in 2020, the FCC approved an order requiring network operators to begin implementing a protocol called STIR/SHAKEN. This protocol allows for one-step backtracking, among other robocall prevention measures.
“One-step traceback is proven,” says Burger. For example, Traceback identified the source of fake Biden calls targeting New Hampshire voters as a Texas-based company called Life Corporation. The problem, Burger says, is that the FCC, the Federal Bureau of Investigation and state agencies do not provide the resources to track so much illegal robocall activity. Historically, the FCC has pursued only the largest offenders.
“There’s no way to stop these calls,” says Hiya’s Nelson. At least not completely. “Our job is not to stop them, but to prevent them from profiting.” Like similar companies, Hiya aims to achieve that goal by lowering the success rate of fraud through protective services that include exposing where calls originate and by whom, thereby reducing the likelihood of individuals answering the calls in the first place.
But Nelson worries that generative AI will lower the barrier to entry so much that these precautions will be less effective. For example, today's scams almost always require the victim to be connected to a live agent at a call center for the scam to be successfully terminated. AI-generated voices could eventually allow fraudsters to shut down call centers entirely.
“Collectively, it is surprising how much money is lost due to scams perpetuated through robocalls.” —Eric Berger (Virginia Tech)
Nelson also worries that as generative AI improves, it will become harder for people to even recognize that they aren't talking to a real person in the first place. “This is where we start to lose ground,” says Nelson. “There may be an increase in the number of call recipients who are completely unaware that it is a scam.” For example, a scammer disguising themselves as a fake charity may successfully solicit “donations” without donors realizing what has actually happened.
“I don’t think you can tell how quickly this will change the phone experience,” says Nelson.
Another issue that complicates enforcement is that the majority of illegal robocalls in the United States originate across the border. For example, Industry Traceback Group found that 65% of all these calls in 2021 were international.
Burger notes that the FCC has taken steps to combat international robocalls. The agency allows other carriers to refuse to carry traffic from gateway providers (a term for network operators that connect domestic infrastructure to international infrastructure) that generate fraudulent calls. For example, in December 2023, the FCC ordered two companies, Solid Double and CallWin, to stop sending illegal robocalls or risk requiring other carriers to reject the traffic.
“I don’t think there is any crackdown. “It’s not the vitality we want to see.” —Eric Berger (Virginia Tech)
The FCC's recent action on AI generating robocalls is a first, and it remains to be seen whether regulators in other countries will follow suit. “I think the FCC is setting a good example by acting quickly and boldly within the scope of its regulatory authority,” Weiner says. But he also pointed out that FCC counterparts in other democracies would likely have more comprehensive results.
According to Burger, it's difficult to say how the FCC's actions will compare to those of other regulators. While the FCC is often ahead, such as when it comes to spectrum sharing, it is often behind when it comes to using mid-band 5G.
Nelson said he expects the FCC's decision to be revised within a few years. That's because it currently prohibits companies from using generative AI for legitimate business practices.
It also remains to be seen whether the FCC's vote will have any real impact. Burger points out that for calls like the fake Biden one, making another aspect of the call illegal would likely not be a game changer, since making a robocall and impersonating the president was already illegal.
“Will triple criminalization really deter people?” Berger says:
From your site article
Related articles on the web