Social Security IG looks to stop fraud revved up by AI

As long as there is Social Security, fraudsters will try to steal it. Now, artificial intelligence is revving up fraudsters efforts. The Social Security Administration‘s Office of Inspector General is searching for ways that it can use AI to stop fraud. For the details, the Federal Drive with Tom Temin talked with the Chief Strategy Officer in Social Security’s Office of Inspector General, Chad Bungard.

Interview Transcript:  

Tom Temin All right. And fraud at Social Security is something I know the agency has been dealing with and is very active about for a long time. Fair to say?

Chad Bungard Absolutely. Since our inception in 1995.

        Join us Apr. 25 at 1 p.m. EST for Federal News Network's StateRAMP Exchange where we'll explore how the StateRAMP program provides cyber assurances as state agencies continue their IT modernization journeys. | Register today!

Tom Temin All right. And in general, what is the way the world stands now? What is the main mechanism for fraudsters at this point?

Chad Bungard So we do report the number of fraud allegations that we receive by category in our semiannual report to Congress. And most recently you’ll see that number one right now are government imposter scams. And, you know, government imposter scams are where someone pretends to be from Social Security or another agency and demands access to information or money. And so, starting in 2018, we saw a huge rise in the number of complaints we received about SSA imposters schemes, peaking in September 2020 with over 110,000 SSA imposters scam complaints. And that is just to our website. I’m not even talking about the numbers that the FTC received. And so quickly we became the number one government imposter scam in the country, even with from the IRS. I think you probably remember, and I received several calls. I even worked from Treasury, and I was getting these calls, and they were scaring me that, you know, from the IRS, but they weren’t from the IRS. And so, then they moved over to SSA, and we quickly became the number one government imposter scheme in the country. And I think we’re still number one, although I’ll touch upon in a second. I think with a lot of our work, those numbers have really gone down. But, you know, at its height, U.S. consumers, many of whom are elderly or otherwise vulnerable, we’re just inundated with millions of illegal robocalls every day. So, I’m sure you’ve received them. I’ve received. Yeah.

Tom Temin In other words, these calls claim to be SSA and get your credentials, and then they can become you with your credentials and get your Social Security payments.

Chad Bungard They’re looking for your information or your money, Tom. So, this is something that we have to tackle quickly and effectively. So, they’re growing an alarming rate and then peaked in September 2020. But we came up with a multifaceted, innovative, multi-pronged approach to combat these damaging scams through public outreach and education, criminal investigation, enforcement and prosecution. But I would say our most successful way of attacking these have probably been through the public outreach and education efforts.

Tom Temin All right.

Chad Bungard Each year we partner with SSA, government agencies and nonprofit organizations and private sector on ‘Slam the Scam.’  Which yesterday Tom, was National Slam the scam day. That’s part of the FTC’s National Consumer Protection Week. That seems like every year we get a Senate resolution unanimously passed declaring national Slam the Scam Day and the whole goal of this, and it’s really taken off. And hashtag Slam the Scam that everybody’s listening. It’s really taking off their awareness for how to spot government imposter scams.

Tom Temin Sure, sure. Let’s get on to artificial intelligence. Now has come a long. And would you say it’s a potential for revving up some of these scams or what’s your concern.

Chad Bungard Absolutely. But I do want to tell the very good story through these education outreach efforts. Tom. And September 2021 as 100 and thousand a month,105,000 a month, we now for the past two years have been averaging maybe 5000 complaints per month, but they’ve gone way down. You’re getting to your artificial intelligence question, right? You probably have noticed that a lot of these calls, there’s broken English. The grammar isn’t the best. They don’t have a good understanding of American culture, but AI is able to fix all that. They are able to run things through. They can run things through ChapGPT and fix all the grammar issues. And so, the bar to entry will be much lower. When it comes to AI manipulated media, there’s no single telltale sign of how to spot a fake. Some of the old-fashioned telltale marks of poor grammar that’s being resolved right now today. And so we are also, I’ve done some presentations for different government agencies on this subject, and one of the things that you’re likely to see more people are likely to see more of, in my view, is the AI generated phishing emails because they look so real and, they can really fool people. So, and then you can have deepfake video where now it takes someone literally who knows what they’re doing. But and I know a guy who learned how to create a deepfake video by watching YouTube tutorials. He created a deepfake video that I demoed in front of a few groups, and he learned it all in two weeks. And now with anybody’s audio and video of just 15 seconds. He can do a deepfake of you, Tom, saying things that you’ve never said before, and that’s scary. You’re also going to see probably an uptick in chat bots, because with AI, they automatically learn it’s not manually done anymore. A few mistakes that they make, they’re able to pick up and how to correct the mistakes that they’re making. But one of the things that I think is always going to remain the same for all of these scams, even in the artificial intelligence world, the thing for people to look at is scare tactics. And I want to remind people of four P’s, which I think are always going to be important even in this artificial intelligence era. And that is scammers pretend, to be from an agency or organization that, you know, to gain someone’s trust. The scammers will say there’s a problem you need to correct immediately. So that’s the second P. There are four P’s here, Tom. And the third P is scammers will pressure you to act immediately. If you don’t call us. And this is an IRS call that I got years ago that scare me. But you you’re behind.  What, I pay my taxes. But it still sounds scary, right? Even the sophisticated and especially the vulnerable. But they pressure you to act immediately, and scammers tell you to pay in a specific way. So those are the things to always look out for, even in this AI world.

        Read more: Benefits

Tom Temin Yeah. Have you seen any of this applied to Social Security related scams yet, or is that just do you feel like, well, it hasn’t happened?

Chad Bungard No, it is hard to say for 100%, certainly, but I believe so. So there are there was a time back in, I would say, 2020, where our 1-800 number in our field offices are receiving calls to the public who sounded very, very real, who we believe, after a careful analysis of some of these calls, that there are very likely to be AI chat bots, and there’s only so much I can say here talking about some of these because of ongoing criminal investigative matters. But I will say some things I can say. At first these callers would call in and they would ask basic questions, right? And then they hung up like they were trying to figure out what information SSA would ask someone when they tried to identify themselves. They’re getting an understanding of questions for information or phishing for information. See how everything work. You’re trying to get a lay of the land, so to speak. So sure enough, these callers again were very likely AI bots began to impersonate beneficiaries or recipients. And run a script with the intent to redirect the direct deposit from the intended victim to the scammer. And some are successful, and there are some hotel signs that I’m unable to get into here. But however, we were able to pin down and stop that in its tracks by working with SSA and SSA addressed this issue through process improvements, but is it possible that it’s still happening and we’re not receiving referrals? Well, yeah, that’s the kind of thing that we would like to dive into.

Tom Temin Well, I guess my question is, what are you mainly recommending to Social Security to be on guard against this, other than to be on guard against it, because it sounds like it takes a particular level of technical understanding to be able to identify these things.

Chad Bungard Yeah. Well, let me say SSA has been a tremendous partner in working on anti-fraud issues. The inspector general sent a letter to the acting commissioner last summer saying, you know, we really need to collaborate together and work together to look at AI related fraud. And also, you know, look at tools to fight fraud, generally with AI. And so, at the same time, the inspector general established an internal task force. And our goal is to study AI and related technology for those two purposes, to fight fraud with AI tools and to fight AI related fraud and SSA, they’ve already had an AI core group. And so, when I say we had a good relationship, we have a very innovative partnership with the agency and it’s very unique in government. We call up the National Anti-fraud Committee. I’m co-chair of that committee with the Deputy Commissioner of SSA, the Office of Analytics, Review and Oversight. And through this, it allows us to raise issues outside of the investigative and audit processes. And we work with the highest levels of the agency to identify real action items aimed at finding, preventing, and detecting fraud. And so, we meet quarterly at the highest levels and then yearly, not only at the highest levels, but we bring in subject matter experts throughout the entire agency. We have like 8 or 9 discrete sections over a three-day period to really tackle issues. In this past year, in September 2023, we held a natural anti-fraud summit focused mostly on AI and the impact of AI on modern fraud schemes. And one of the action items that came out of that summit last year was to hold quarterly meetings with our AI task force, OIG, and the agency’s AI core team to discuss all of the different issues, and they’re going to be transparent. We’re all going to be transparent. We’re going to be working together to fight fraud. Our first meeting on that is March 20th. The bottom line, to answer your question more directly here is, I think, and this is for every agency, not just SSA. I think all agencies need to collaborate with their respective OIG first and foremost, and then all agencies need to collaborate with one another. Yeah. When we’re facing questions related to artificial intelligence, the inspector general and I talked about. And she said to me Chad, why would we be fighting this alone? Other agencies surely are facing the same type of issues or will be facing the same type of issues that we’ll be facing. So, I reached out to OMB, and they were phenomenal. They invited me to two chief Financial Officer symposiums where we talked about AI and fraud. And just this past week on Tuesday, I spoke at the CIO symposium with chief CIO for the white House, Clare Montorano, and the acting comptroller Deidre Harrison and AG, a statistician, where we talked about AI and fraud. So, I think number one initiative to collaborate with their respective OIG to maybe collaborate with one another. We can share best practices, we can collaborate, we can share, hey, the trends that they’re seeing there, they’re seeing a trend, you know, as likely to come our way potentially. And then, you know, third thing I would recommend is implement an AI strategic plan. There are a bunch of things in here that I would include like AI algorithms to analyze vast amounts of data in real time to identify patterns, anomalies, and suspicious activities indicative of fraud. Two, I would also implement AI powered behavioral analysis so we can monitor to detect unusual behavior patterns associated with fraud, which SSA has done in some form or fashion. With AI, there’s so much more probably can be done. Third, utilize AI tools for real time monitoring. So, as it occurs, real time alerts and notifications are given to the frontline staff so we can able self-response to mitigate risk at SSA. And then fourth thing that needs to be in that strategic plan. And I could you the big list, but I’ll stop here is develop the AI based fraud prevention mechanisms that proactively identify and block fraudulent activities before they occur. SSA is doing this to some extent. It can be improved upon through AI, but this can include setting out, you know, rules-based systems, anomaly detection, algorithms that goes a hand can work together with predictive modeling.

Copyright © 2024 Federal News Network. All rights reserved. This website is not intended for users located within the European Economic Area.