As humans, we’re hard-wired to find comfort in the voices of people we know. Research suggests that certain voices, like those of a parent or partner, elicit strong neurological and physiological responses, from lighting up the parts of our brain associated with reward to boosting oxytocin and lowering stress hormones.
Nearly 150 years ago, when the telephone was invented, it felt like nothing short of a miracle that people could get that warm, fuzzy reassurance by listening to the voices of loved ones hundreds of miles away. Today, it’s a phenomenon we take for granted, with voice notes and video calls a routine form of everyday communication.
Voice communications like this have also historically felt like relatively safe territory in the context of fraud: If you heard and recognised a familiar voice on the other end of the line, you had little reason to question it. Even as phishing and text-message scams like smishing became more advanced throughout the 2010s, a loved one’s speaking cadence or facial expressions still felt difficult to counterfeit.
But today, that’s no longer the case. AI voice-cloning technology has become scarily convincing in just the past few years, and it seems to improve with every new model release. Video, too, has come leaps and bounds from the first “Will Smith eating spaghetti” viral video of 2023.
Those advances have opened up a powerful and disconcerting new avenue for fraud: AI deepfake scams.
Why AI Voice and Video Scams Are So Convincing
Voice is only one element of the most sophisticated scams today. Bad actors often build their impersonations with details about your loved one’s life or job gathered from <a href="https://aikporia.kporia.com/9-canva-ai-features-for-social-media/" title="9 Canva AI Features for Social Media“>social media or past data breaches. So, they may know the name of your boss, the country your daughter is currently travelling around, or which bank branch you visited yesterday. Caller ID can also be spoofed, so even a familiar number may not be all it seems.
Those little details, combined with a voice engineered to sound remarkably like somebody you trust, can make a scam call, video, or voice message feel bona fide. Even if something seems slightly amiss, the verisimilitude of the person on the other end can make instincts such as “I’d know their voice anywhere” much less useful.
Red Flag No. 1: A Sudden Sense of Urgency
Despite these realities, there are a few red flags that can give away the game: An urgent, unexpected request is perhaps the most prominent “tell” of potential fraud, whether it arrives in the form of a phishing email or a panicked phone call.
This can take a few forms: Your daughter might claim to have been arrested abroad and need money for bail, or a manager might insist that a bank transfer go through at 4:45 p.m. for a client contract due EOP. Whatever the story, be cautious around requests with “you must act immediately” undertones.
Secrecy can serve a similar purpose. Be particularly wary if a caller insists you shouldn’t tell another family member or colleague what’s happening, or becomes agitated when you suggest hanging up and calling them back.
The other obvious warning sign is being asked for money or sensitive information off the bat. If someone is asking you to transfer them funds or provide login credentials or security codes, that merits an independent check, even if every syllable sounds exactly like your son or work superior.
Red Flag No. 2: Something Doesn’t Quite Add Up
Even though voice cloning is improving quickly, some synthetic voices still have quirks: you might notice an unnatural speaking rhythm, or an emotional delivery that doesn’t quite match the topic of conversation. Video deepfakes add another set of potential giveaways: audio and lip movements may not align perfectly, or details around the face may distort as the person moves.
Still, these clues are becoming less dependable as the technology improves. Not to mention, there are valid reasons for technical glitches: a poor Wi-Fi connection or patchy mobile signal, for example. But especially when combined with the messaging tells described above, audio and visual hiccups should give you extra pause.
Verify the Person Somewhere Else
The single most important step a potential deepfake victim can take is to maintain enough presence of mind to verify the request separately.
That is, of course, easier said than done when someone who sounds exactly like your daughter is sobbing down the phone. A scammer may even beg you not to hang up, claiming their phone is about to die or that there isn’t time to call someone else.
Hang up anyway. Ring your loved one using the number already saved in your contacts, rather than a number provided during the suspicious call. If they don’t answer, try somebody else who is likely to know where they are. For a work-related request, verify an unusual payment request through your normal internal channels, especially if there’s a financial element involved.
This is also where a predetermined family code word or question can come in handy. Agree on something that only your loved one is likely to know, and ask for it if a family member supposedly calls with an emergency. Choose something that’s both obscure enough that it can’t be easilyvent of a stressful situation
Add Another Layer of Protection
The habits above are most helpful when you’re already being actively targeted during a phishing attack, but certain cybersecurity tools can help reduce the chances of getting that far.
Bitdefender Mobile Security for Android, for example, includes Call Blocking, which can warn you about or automatically block numbers associated with scams and spam. (Bitdefender also offers a free Reverse Phone Lookup tool that lets you check whether an unfamiliar number has previously been associated with suspicious activity.)
Bitdefender Scamio is another line of defence if you’ve received a suspicious-looking text or email. If you receive a fishy payment link or “off-seeming” screenshot, for example, you can share that content with Scamio to analyse before engaging with it.
While AI doesn’t mean we need to distrust every call from mum, it does mean that hearing their voice is no longer quite the reassuring form of proof it once was. Staying in tune with the latest threat landscape and implementing some proactive measures can help keep that all-too-believable message from becoming a costly mistake.
ExploreBitdefender’s range of security solutionsto protect yourself from AI-powered phishing scams and other forms of fraud.
