Deepfake Voice Scams: How to Verify Your Loved Ones

Imagine picking up the phone to hear your child or spouse panicking, claiming they have been arrested, injured, or kidnapped. The voice sounds exactly like them, capturing their specific inflection and tone. In the past, this was a clear sign of an emergency. Today, it is increasingly likely to be a scam fueled by artificial intelligence. With audio cloning technology becoming cheaper and more accessible, families must adopt new strategies to verify who is actually on the other end of the line.

The Rise of the AI "Grandparent Scam"

The traditional “grandparent scam” involved a fraudster calling an elderly person, pretending to be a grandchild in trouble, and asking for money. These used to be easy to spot because the voice never sounded quite right. The scammer would claim they had a cold or a bad connection.

AI has eliminated that barrier. Scammers now use voice-cloning software to create a near-perfect replica of a loved one’s voice. According to a 2023 study by McAfee, 70% of people surveyed were unsure they could distinguish between a cloned voice and a real one. Even more concerning, the Federal Trade Commission (FTC) reported that impostor scams resulted in $2.7 billion in losses in 2023 alone.

How the Technology Works

You might think scammers need hours of recordings to clone a voice. In reality, they need very little data. Sophisticated AI models can generate a convincing clone with just three seconds of audio.

Scammers harvest this audio from public sources:

  • Social Media: Videos posted on TikTok, Instagram Stories, or Facebook reels often contain enough clear speech to train an AI.
  • Voicemail Greetings: If your voicemail includes you saying your name and a greeting, that is often sufficient.
  • Spam Calls: Sometimes, a silent call that prompts you to say “Hello?” or “Who is this?” is actually recording your voice for future cloning.

Once the clone is created, the scammer types text into a program, and the AI reads it aloud in your voice. They can even add emotional layers, making the voice sound fearful, crying, or desperate.

The Jennifer DeStefano Case

To understand the severity of this threat, look at the widely reported case of Jennifer DeStefano from Arizona. In 2023, she received a call from an unknown number. When she answered, she heard her 15-year-old daughter sobbing. A man then came on the line, threatening to harm the girl unless DeStefano paid a $1 million ransom.

The voice was identical to her daughter’s. It was not until DeStefano was able to contact her husband and confirm their daughter was safe at a ski trip that she realized it was a hoax. The trauma, however, was real. This incident highlights why relying on your ears is no longer a safe verification method.

The "Safe Word" Strategy

The most effective low-tech defense against high-tech fraud is a family “safe word” or “code word.” This is a pre-agreed word or phrase that family members must provide to prove their identity during a crisis.

How to Implement the Code Word

  1. Hold a Family Meeting: Sit down with your spouse, children, and close relatives (like grandparents) to discuss voice AI. Explain that voices can be faked.
  2. Choose a Unique Word: The word should be easy to remember but not something you would naturally say in a panic. Avoid common passwords or birthdates.
    • Good Examples: “Purple Monkey,” “Jupiter,” or a specific obscure vegetable like “Rutabaga.”
    • Bad Examples: “Help,” “Password,” or the name of the family dog.
  3. Practice the Scenario: Tell your children: “If you are ever in trouble and call me from a strange number, I will ask for the code word. You must say it immediately.”
  4. Use It Ruthlessly: If you receive a call claiming your loved one has been kidnapped or arrested, ask: “What is the code word?” If the caller gets aggressive, makes excuses, or hangs up, it is a scam.

Additional Verification Tactics

If you haven’t set up a code word yet, or if the caller claims they are a police officer holding your family member, use these verification steps:

The “Hang Up and Call Back” Rule

This is the golden rule of telecom safety. If someone calls claiming to be your child from an unknown number, hang up immediately. Then, call your child’s actual mobile number. If they answer and are safe, the previous call was a scam. Scammers often claim, “I only have one call,” or “They took my phone,” to prevent you from doing this. Ignore those threats.

Verify with Other Family Members

If you cannot reach the person supposedly in trouble, call their spouse, roommate, or school. In the DeStefano case, a call to her husband confirmed the daughter was safe within minutes.

Ask “Proof of Life” Questions

If you are kept on the line, ask questions that an AI or a stranger who browsed your Facebook profile would not know.

  • Ask: “What did we have for dinner last night?”
  • Ask: “What represents the ‘blue’ on your bedroom wall?”
  • Do Not Ask: “What is your dog’s name?” or “When is your birthday?” (This information is easily found online).

Strengthening Your Digital Privacy

Prevention is better than verification. You can make it harder for scammers to clone your voice in the first place by locking down your audio footprint.

  • Audit Social Media: set your Instagram and Facebook profiles to “Private.” If your accounts are public, anyone can download your videos and clone your voice.
  • Limit Biomarker Sharing: Be cautious about participating in viral social media trends that ask you to “record your voice” or answer questions on video.
  • Voicemail Security: Consider removing your actual voice from your voicemail greeting and using the automated system voice instead. This removes a primary source of clean audio for scammers.

Frequently Asked Questions

Can the police trace these calls? It is very difficult. Scammers use “spoofing” technology to make the call look like it is coming from a local area code or even the loved one’s actual contact number. They also use VPNs and operate from overseas jurisdictions, making prosecution challenging for local law enforcement.

Is voice cloning illegal? Using voice cloning for fraud is illegal. In February 2024, the Federal Communications Commission (FCC) officially declared that robocalls made with AI-generated voices are illegal under the Telephone Consumer Protection Act. This gives state attorneys general more power to prosecute the companies that facilitate these calls.

Does this only happen to wealthy people? No. These are volume-based crimes. Scammers use automated dialers to call thousands of people. They do not necessarily target you because of your net worth; they target you because you answered the phone.

How much does voice cloning cost the scammer? It is incredibly cheap. Subscriptions to high-quality voice AI services like ElevenLabs can cost as little as $5 to $20 a month. While legitimate companies try to ban bad actors, scammers often find workarounds or use open-source software like Microsoft’s VALL-E research models that have been leaked or replicated.