For years, some of the simplest personal safety advice was built around what we could see and hear.
Does the person look like their profile picture? Does that voice actually sound like your son, your friend, or your boss? Does the message read like something they would normally send?
Artificial intelligence is making those questions much harder to answer.
We have entered a period where a convincing photograph, a familiar voice, a polished message, or even a video call can no longer be treated as proof that the person on the other side is who they claim to be. That has implications far beyond cybersecurity. I believe it is becoming a personal safety issue.
The numbers are already significant. In its 2025 Internet Crime Report, the FBI’s Internet Crime Complaint Center recorded 22,364 complaints containing AI-related information, with adjusted reported losses of more than $893 million. The FBI also reported more than $19 million in losses from confidence and romance scams with a likely AI connection, while AI-related investment scam complaints accounted for more than $632 million in reported losses.
There is an important caveat to those numbers. The FBI defines its “AI Related” descriptor as a complaint containing a reference to artificial intelligence. It is not a forensic determination that AI caused every dollar of the reported loss. That distinction matters, especially when discussing a technology that is still evolving quickly.
But the underlying problem is difficult to ignore.
The scams aren’t necessarily new. The tools are.
Impersonation is hardly a new form of fraud. Criminals have pretended to be relatives, government officials, romantic partners, banks, employers and celebrities for generations.
What AI changes is the quality and scale of the deception.
The FBI has warned that generative AI can help criminals produce believable social media profiles, realistic images, convincing messages and synthetic audio while reducing the time and effort required to create them. It has specifically documented the use of AI-generated images for fake profiles, cloned audio to impersonate relatives, and synthetic video that can be used during online conversations to make a fictitious person appear real.
That matters because many of the habits we developed for spotting scams were based on imperfections.
We were told to look for awkward grammar. Strange photographs. A voice that doesn’t sound quite right. A person who refuses to video chat.
Those clues can still be useful, but we shouldn’t depend on them nearly as much as we once did.
In 2025, the FBI even warned about a campaign in which malicious actors used text messages and AI-generated voice messages while impersonating senior U.S. officials. The goal was to establish trust and ultimately gain access to accounts or sensitive information.
At the same time, the broader impersonation problem continues to grow. The Federal Trade Commission says consumers reported losing $3.5 billion to imposter scams in 2025, and nearly one in three fraud reports that year involved an imposter scam. Those figures cover impersonation fraud generally, not AI specifically, but they illustrate the size of the environment into which generative AI is being introduced.
We need to rethink what “verification” means
This is where I think personal safety education needs to evolve.
The answer cannot simply be, “Get better at spotting deepfakes.”
Technology will keep improving. Asking the average person to become an expert in synthetic media detection isn’t a realistic long-term safety strategy.
Instead, we need to move toward what I think of as independent verification.
Imagine receiving a call from someone who sounds exactly like a family member. They’re frightened. They say they’ve been arrested, injured or stranded somewhere and need money immediately.
The natural instinct is to focus on the voice: That sounds like them.
The safer question is: Can I confirm this another way?
The FTC has advised consumers confronted with this kind of family-emergency scenario not to rely on the voice alone. Instead, contact the person directly using a phone number you already know or reach another trusted person who can verify the situation.
The FBI similarly recommends independently verifying someone’s identity and even establishing a private word or phrase among family members that can be used when identity is in doubt.
That is a small behavioral change, but an important one.
Slow down when someone wants you to speed up
There is another element of these scams that technology hasn’t changed very much: human psychology.
Urgency remains powerful.
Send the money now. Click this link immediately. Don’t tell anyone. Your account is about to be closed. Your daughter is in trouble. Move the conversation to another platform.
The more sophisticated the technology becomes, the more important it is to pay attention to the behavior surrounding the request rather than just the quality of the content.
A perfect-looking video doesn’t make an unusual financial request reasonable. A familiar voice doesn’t mean you should surrender an authentication code. A polished dating profile doesn’t mean you should ignore inconsistencies in someone’s story.
Sometimes the safest move is remarkably low-tech: stop the conversation, step away from the emotional pressure and verify what you have been told through another channel.
Online safety and physical safety are merging
At SwypeSure, we spend a lot of time thinking about personal safety as something broader than what happens on a street or in a parking lot.
Increasingly, an interaction can begin online and eventually affect someone’s finances, privacy or physical safety.
A dating-app conversation can turn into an in-person meeting. A social media connection can become a financial relationship. Someone who appears trustworthy online may eventually learn where you work, where you live or how to contact the people closest to you.
That is why AI-powered impersonation shouldn’t be treated as solely a technology problem.
It’s becoming part of everyday safety.
This doesn’t mean we should stop trusting people, abandon online dating or assume every unusual message is generated by AI. Fear isn’t a useful personal safety strategy either.
It does mean that trust may increasingly need an extra step.
The lesson of the AI era may ultimately be surprisingly simple: seeing is no longer always believing, and neither is hearing.
We are going to need to become more comfortable verifying before trusting, slowing down before reacting and recognizing that the most convincing part of an interaction may no longer be the most reliable.
Technology has changed what can be faked.
Our safety habits need to change with it.
