You get a call late one afternoon that no parent wants: “Dad, I crashed the Toyota, and someone got really hurt. I need $10,000 for bail money today. I’m so sorry. I’m putting the officer on the line about wiring money….”
The voice and personal details are definitely your daughter’s—but the call is actually a high-tech scam. The perpetrators used artificial intelligence (AI) to replicate her voice and deploy it against her family.
You may think you’re savvy about fraudulent calls and texts, but AI is transforming the way these crimes occur. Instead of poorly produced fake e-mails full of typos and awkward phrases, the attacks have become far more sophisticated and convincing.
Con artists can create realistic voices, websites and video clips to mislead consumers, rip off investors and “phish” for personal information. According to Deloitte’s Center for Financial Services, fraud losses enabled by generative AI in the US are expected to reach $40 billion by 2027.
Fraud-prevention expert Frank McKenna says you don’t need to be a cyber expert to avoid getting victimized. You can use low-tech tactics to unmask and foil even the most convincing AI-empowered impostors. Here are the latest AI scams and McKenna’s clever strategies to protect yourself…
Scam: AI Voice Cloning
How it works: AI technology can mimic a real person’s voice with 85% accuracy from just a three-second sampling of audio, according to McAfee security researchers. Typically, the con artist collects the audio data from a person’s social-media pages (e.g., Facebook, Instagram), then feeds it into AI software. The software analyzes the voice’s unique characteristics such as pitch, tone and rhythm to generate a synthetic audio of that person that can interact in real time. The cloned voices are used to impersonate a loved one in distress or conduct fake business transactions. It is even possible for the scammer to respond in real time with voice-changing AI software. Example: You are in the process of making a down payment on a house, and you get a call from someone who sounds like your escrow agent instructing you to wire your payment to the criminal’s account.
Self-defense…
- Listen for telltale signs of an artificially engineered voice. These include the lack of breaths and moisture sounds between phrases that humans make as they inhale and open and close their mouths…monotone speech patterns…abrupt or cut-off endings of sentences.
- Establish a family code word. It should be a word or phrase that is unique and not used in regular conversation. The code word can be used in calls about emergencies or financial requests so the family knows the person calling is really who they say they are. If the caller can’t name the word, hang up. Also consider having a predetermined security question that can be asked even after the code word is received.
- Use the default automated-voice greeting on your smartphone voicemail. Reason: If you record a greeting in your own voice, con artists can easily use it to clone your voice.
Scam: AI Data Sniping
How it works: A text message pops up on your phone with a photo of your vehicle with your license plate parked in front of a local strip club. “Let’s play a game. YOU have been a bad little boy,” the text reads. “Nobody has to see this picture. It would be a shame if it was sent to you know who. Send $1,000 by midnight tonight, and it will go away quietly and you will never hear from us again.” You probably have never been to that club in your life, but an AI program has scraped enough public photos to make the extortion threat feel very credible.
Self-defense…
- Don’t pay ransoms. Paying the ransom usually doesn’t eliminate the threat, and those who do pay often are approached for more money or get retargeted later. Better: Reduce the criminals’ leverage. Their greatest power is your emotional reaction, so don’t engage. Don’t beg, argue or threaten. Silence removes their control. Most of these attacks are volume-based…and the scammers move on to easier targets when leverage drops.
- Lock down your accounts in case the threat is real. Change your passwords. Enable two-factor authorization on all accounts. Revoke third-party app access. Scan your devices for malware if anything was shared from them
- Preserve evidence and report it—even it feels awkward. Inform the security department for the platform where the threat came from (WhatsApp, e-mail providers, etc.). Create and save screenshots of all the messages…usernames…wallet addresses…e-mails…headers…file names…dates and times. You also can file a complaint at the FBI’s Internet Crime Center at ic3.gov.
- Confide what’s going on with someone you trust such as a friend or sibling. Keeping secrets raises anxiety and panic levels. Shame loses some of its power when you’re not alone with it.
- Plan a calm response in the rare event that a criminal does send embarrassing material to someone you know. The imagined fallout often is much worse than reality. People are more compassionate than you would expect. Ask the recipient not to open the e-mail or forward it. Simply say, “Someone is attempting to harass me using stolen material. It’s being handled.”
Scam: AI Deep-Fake Celebrity Videos
How it works: Chef Gordon Ramsay endorses cookware…Tom Hanks recommends dental insurance plans…Elon Musk promises guaranteed returns on cryptocurrency investments. These fake ads all have appeared on Facebook, TikTok, YouTube and other platforms. But they’re AI imitations of real people, generated from pre-existing audio, videos and photos. AI software studies a celebrity’s head and facial-muscle movements, then maps the celebrity’s face onto someone else’s performance.
Self-defense…
- Look for “tells.” If something seems off, but you don’t quite know what it is, trust your instincts. AI technology still struggles to replicate certain features, including facial hair, which often looks unnatural…blinking too little or too rhythmically… lips slightly out of sync with audio…reflections in eyes that don’t match the scene lighting…shadows shifting unnaturally when the head moves.
- Google the celebrity’s name and product/service to check if the ad has been called out as a fake or a scam.
Scam: AI Search Poisoning
How it works: You want to plan for a vacation, so you do what millions of people now routinely do every day—go to an AI search engine such as ChatGPT, Google Gemini or Claude, and ask for recommendations, itineraries and reservations. You call one of the travel customer-service numbers and book dates with an agent. But the number and agent turn out to be part of a scam call center. A string of fraudulent charges soon hit your credit card. Scammers are flooding the Internet with fake hotline and customer service numbers on hacked websites, fake YouTube videos and manipulated Yelp reviews.
Self-defense…
- Always book directly through well-known official websites or trusted travel platforms. At the very least, cross-check booking details by logging into the airline or hotel’s official website.
- Save official airline contact numbers before you fly. If your flight gets canceled, the urge to quickly use ChatGPT to help book a flight is what scammers count on.
Scam: AI Spoofing of Financial Advisors
How it works: A financial advisor receives communication from a criminal pretending to be one of their clients and asking to withdraw $100,000 in funds for a home renovation to be wired to a specific account. The advisor sends out permission documents to the real client to be signed electronically and authorizing the withdrawal and wiring instructions. The client fails to read the documents closely and signs off on them. After the scam is discovered, the advisor claims he/she isn’t at fault since the client approved the documents.
Self-defense…
- Ask your financial advisors about their cybersecurity policy and protocols. Make sure no major or unusual transactions can be carried out without speaking directly with you first. Transactions should not be given the go-ahead solely through e-mail or voicemail.
- Pursue action against your financial advisor if you do get scammed this way. Compensation is possible if the advisor failed to act with reasonable care or violated industry rules. You can seek recovery through FINRA arbitration (for licensed brokers) or civil lawsuits for negligence or breach of fiduciary duty.
