The evolution of artificial intelligence (AI) has reached a point where many apps can mimic not only human behavior in general, but human images and voices as well—possibly even those of your friends and family members.
One result of this innovation is that scammers are now using this so-called “deepfake AI” to commit identity theft and steal banking and credit card information from victims.
Many sophisticated AI deepfakes can converse realistically. And the quality of their voices and video images is often difficult to distinguish from the real thing, making these scams difficult to detect.
How it's done
Scammers typically gather audio by hacking voicemail systems and podcasts and capturing publicly available video of people on social media.
According to research from McAfee, AI deepfake apps only need a few seconds of a recorded voice to create an impersonation that is 85% accurate.
Using deepfakes in scams
Typically, a victim receives a voicemail or video message from someone who convincingly sounds or looks like someone you know. The scammer may pretend to be your spouse, asking you to provide your Social Security number for an application they’re filling out.
Or they may be impersonating one of your children or grandchildren, requesting money to be wired to them to bail them out of a financial emergency.
How prevalent are these scams? According to McAfee, one in ten people surveyed have received a message from an AI voice clone, and 77% of them lost money as a result.
Fending off the fakes
No one wants to hang up on or refuse to help a loved one who may be facing a genuine emergency.
At the same time, people need to be more vigilant than ever against the growing power of AI deepfakes—or any kind of scam attempts from criminals posing as people you know.
Fortunately, there are steps you can take to make sure these kinds of requests are genuine.
- If you receive any kind of audio or video message from someone you know asking for money or personal information, don't commit immediately. Contact that person to see if the request was genuine. Chances are, it wasn’t.
- Use “safe words.” Give trusted friends and family members certain words or phrases that only they and you know. If you ever have a live conversation with someone you believe is an AI impersonator, ask them for the safe word. If they can’t provide it, hang up immediately.
- A similar strategy is to use “password pairs” that you and other people both agree on. Uttering them in sequence confirms the authenticity of both people. For example, when someone calls you, their safeword might be “Boston” and your follow-up might be “Baked beans.”
- If you haven’t set up safewords with someone you think may be an impersonator, ask them to confirm their identity by describing a memory of an event or conversation only the two of you would remember.
If you’ve foiled one of these deepfake attempts, immediately contact the impersonated friend or family member to let them know that someone may have stolen their voice or image…and possibly more.
This article was authored by David Jaeger and Jeffrey Briskin. David is a financial advisor with Canby Financial Advisors, a SEC-registered investment adviser. SEC registration does not constitute an endorsement by the SEC nor a statement about any skill or training. David can be reached at 508.598.1082 or djaeger@canbyfinancial.com. Jeffrey Briskin is Director of Marketing at Canby Financial Advisors.
©2025 Canby Financial Advisors.