Yesterday I was letting my phone play me random podcasts while I lifted weights in my basement, and I heard a short episode (9 minutes) of The Indicator from Planet Money that set me thinking.
The Indicator is a spinoff of shorts from the main Planet Money economics podcast. This week they’re doing a series on financial crime, data breaches, etc. in the current technological environment. The episode I heard yesterday was about AI deepfakes:
REPORTER: A lot of people are falling for these kinds of audio deepfakes. It’s like millions of Americans have lost money to a scam call that uses an AI voice. And the losses from these scams can be in the thousands of dollars.
Banks are a big target of AI voice fraud… One scheme calls people up, it records you talking for several seconds, and then it turns that into a cloned AI voice, and uses that to bypass banks’ voice verification on the phone.
The episode starts out with audio of one of the show’s reporters using a deepfake of his own voice —at least that’s what he tells us he’s doing—to try to trick a colleague. The colleague doesn’t fall for it, but assuming that he really is using an AI-generated voice clone, we can hear that the voice is pretty close.
Audio voice impersonation is an element in crimes ranging from corporate bank theft to faked personal endorsements of “bogus investment schemes” to tricking individuals into thinking that a loved one has called begging for money to be wired to them, according to this recent article from the American Bar Association.
The information presented in the episode would imply that there is considerable risk in speaking on the phone to an unknown caller, unwittingly providing a voice sample that could be cloned.
The longer the sample, the better the clone can be; the episode didn’t mention this, but long and detailed voice (and video) samples could even more easily be lifted from video that people intentionally post to social media, no spoofing required. And of course there’s plenty of that out there; perhaps the phone is a comparatively small risk.
Still, there are plenty of folks out there who haven’t made any social media videos, and for whom the phone is the most likely place that their voice could be captured. Some of them control a lot of money, and others don’t particularly have a lot but would be anxious to send what they can to help what sounds like a weeping, frantic grandchild.
Eventually, technology and regulations may give us ways to protect ourselves by detecting clues in the audio signature of cloned voices. Banks and other financial companies are already doing so. The podcast quotes an owner of a company that makes detection software for banks; he advocates (unsurprisingly) for “all content online to be vetted for whether it was AI generated from text to voice to video”:
SOFTWARE EXECUTIVE: I think we’re going to look back and say, “We can’t believe there was a time when we didn’t have automated deepfake detection.” Our challenge is that technology is moving quicker than regulations.
For the time being, he and other authorities recommend establishing a “safe word” with family members that can be used to establish that they are who they say they are should they ever (apparently) call out of the blue and ask for money. I’m used to this already: I’ve been known to check the identity of my own adult kids when they text me and ask for, say, the Amazon password, either by dialing their number to initiate my own voice call or by asking for a bit of shared info that strangers wouldn’t easily guess.
Note: I’ve no idea whether they think this is cringe, Savvy Old Person Behavior, or just common sense; honestly it doesn’t matter. I’ve noticed a slightly disturbing trend of my younger teenage kid volunteering information to me about Internet Scams He Heard Are Targeting The Elderly. I’m working hard to answer positively (and not, say, with “How new to the Internet do you think I am? Do I look like someone who would meet a ‘CIA agent’ on the sidewalk to give him $50,000 cash in a shoebox?“)
But it’s also got me thinking about how I answer the phone. I get a lot of calls from unknown numbers. My iPhone helpfully marks some of them “Scam Likely” (as in the above photos) and increasingly, “Charity Call.” Unfortunately, it seems that occasionally calls are misdiagnosed: for example, the box office of a theatre where I have season tickets often shows up as “Charity Call,” and my kids’ healthcare providers have occasionally pinged as “Scam Likely” if I haven’t gotten around to adding as contacts. So sometimes I pick those calls up anyway, although it would really be wiser to let just about everything go to voice mail.
I suppose there are probably still a lot of folks out there who would gladly pick up a “Charity Call” and listen to the pitch, and perhaps ask questions, and consider donating to whatever charity was on the line. Giving to charity is a strong value for many, and Mt 5:42 has a broad command: “Give to the one who asks you”; you can parse it carefully and note that it doesn’t say give everyone what they ask for, but the scrupulous might consider it binding.
But increasingly we can’t be sure that the “charity call” is what it says it is (I myself am constantly getting calls from different numbers but the same distinctive voice asking for money for a national lobbying organization that pretends to be a charity and is definitely not, no matter how many times I tell him not to call me) and, it turns out, it could even be dangerous to speak for more than a moment to the callers. Social media is already overrun with donation requests from unvetted sources making highly emotional claims. This last bit is hard to accept, especially since we know that legitimate requests for help from honest people also come to us that way, verified by our own personal networks.
Prudence may, in the end, require us to stop accepting unknown calls at all, and generally to refuse requests for charities that come over the phone or internet in almost any form. But that’s going to be hard for a lot of generous people to swallow, and it does raise the question of when vigilance becomes an excuse to never give at all, or how not to forget about the people who don’t have another way to beg.
bearing blog 