The Call You Didn't Answer (And the One You Did)

Sarah stared at the unknown number on her screen. She let it ring.

Her 78-year-old mother, Margaret, didn't.

The voice on the other end of Margaret's call sounded exactly like Sarah—panicked, tearful, desperate. There had been an accident. She needed $8,000 wired immediately for emergency surgery. Please, Mom. Please.

Margaret's hands shook as she drove to the bank. Within an hour, the money was gone.

Sarah called that evening. "Mom, I've been at work all day. What accident?"

The voice on the phone hadn't been Sarah. It had been an AI—trained on clips from Sarah's Facebook videos—recreating her daughter's voice with frightening accuracy. The scammer needed less than 30 seconds of audio to clone it.

This story plays out 56 million times every year in America. And it begins with a simple, everyday decision: whether to answer an unknown call.

The One-in-Three Problem

Here's what most people don't realize: when an unknown number appears on your phone, there's roughly a one-in-three chance it's spam or fraud.

Think about that for a moment. Every third call from a number you don't recognize is either a waste of your time or a potential threat.

Best case scenario? It's spam. Someone trying to sell you an extended car warranty, a Medicare plan you don't need, or a vacation you didn't win. You lose 4 minutes of your life—the average length of these calls—plus the mental disruption of context-switching from whatever you were doing.

Worst case scenario? It's a scam. And if you're one of the 16% of Americans who lost money to phone fraud last year, you're looking at an average loss of $2,257. That's nearly four times what the average American spends monthly on food.

In any case, it's costly. Americans collectively spend over 7.6 hours per year just screening unwanted calls. Multiply that by the national workforce and you're looking at billions in lost productivity.

The Explosion of AI-Powered Threats

What changed? One word: AI.

The 2025 CrowdStrike Global Threat Report revealed that vishing (voice phishing) attacks increased 442% from the first half of 2024 to the second half. That's not a typo. Four hundred forty-two percent.

The technology enabling these attacks has crossed what researchers call the "indistinguishable threshold." Voice cloning tools can now recreate a convincing replica of someone's voice from just a few seconds of audio—complete with natural intonation, rhythm, emphasis, emotion, pauses, and breathing patterns.

The Arup case illustrates just how sophisticated these attacks have become. In early 2024, an employee at the multinational engineering firm Arup was invited to a video call with what appeared to be the company's CFO and other senior executives. Following the CFO's instructions, the employee transferred $25 million across 15 transactions to bank accounts in Hong Kong.

Every single person on that video call was a deepfake.

"What we have seen is that the number and sophistication of these attacks has been rising sharply in recent months," Arup's Chief Information Officer Rob Greig told the World Economic Forum. "This happens more frequently than a lot of people realize."

According to forecasts from the Deloitte Center for Financial Services, fraud losses facilitated by generative AI are projected to climb from $12.3 billion in 2023 to $40 billion by 2027—a compound annual growth rate of 32%.

The Scale of the Spam Epidemic

Beyond the fraud, there's the daily deluge of unwanted calls that has made the telephone—once a lifeline of communication—into a source of anxiety.

According to the YouMail Robocall Index, Americans received:

• 52.5 billion robocalls in 2025

• 2.56 billion spam and telemarketing calls per month (up 20% from 2024)

• An average of 14 spam calls per person per month

That's roughly 140 million robocalls every single day—or 1,627 calls every second.

The result? People have stopped answering their phones. This creates a costly paradox: the spam problem is so severe that people miss important calls from doctors, schools, and employers because they've been conditioned to ignore unknown numbers.

Why Your Phone Line Is the Last Unprotected Door

We've become remarkably good at protecting ourselves online. Password managers. Two-factor authentication. Email filters that catch 99.9% of phishing attempts. We've trained ourselves to be skeptical of written messages from unknown senders.

But when the phone rings? We still answer.

Understanding the threat means understanding its evolution. Today's phone-based attacks bear little resemblance to the obvious cons of a decade ago.

The AI Impersonation Call

This is what happened to Margaret. Scammers use AI to clone the voice of someone the victim knows—a child, grandchild, spouse, or close friend. They create an emergency: an accident, an arrest, a medical crisis. The emotional urgency bypasses rational thinking.

Variations include:

• The grandparent scam: "Grandma, I'm in trouble and I need help"

• The kidnapping scam: "We have your daughter. Don't call the police"

• The accident scam: "I'm in the hospital. Please send money for treatment"

According to McAfee research, 40% of people said they would help if they received a voicemail from their spouse who needed assistance. And one in ten people report having received a cloned voice message—77% of whom lost money.

The Authority Figure Scam

Scammers pose as representatives from trusted institutions: the IRS, Social Security Administration, Medicare, or local law enforcement. They claim there's suspicious activity, a warrant for arrest, or benefits at risk.

The goal is to create fear of governmental authority combined with a clear "solution"—usually payment via gift cards, wire transfer, or cryptocurrency.

The Tech Support Scam

A caller claims to be from Microsoft, Apple, or your internet provider. They've detected a problem with your computer. They need remote access to fix it—or payment to prevent further damage.

According to the FBI IC3, tech support fraud resulted in $1.46 billion in losses in 2024, with people over 60 accounting for the majority of victims.

The Investment Scam

Particularly targeting retirees, these calls promise high-return investments, often involving cryptocurrency. The scammer builds trust over multiple calls before requesting larger sums.

The FBI calls this "pig butchering"—fattening the victim before the slaughter. Investment fraud was the costliest category in 2024 at $6.57 billion in reported losses.

What You Can Do Today

While comprehensive phone protection is still emerging, there are steps you can take right now to reduce your risk—and protect the people you love.

Establish a Family Code Word

Choose a word or phrase that only your immediate family knows. If anyone calls claiming to be a family member in distress, ask for the code word. A scammer—even one with a cloned voice—won't know it.

This simple technique would have saved Margaret $8,000.

Verify Before Acting

If you receive an urgent call from a bank, government agency, or any organization requesting immediate action or payment:

1. Hang up

2. Find the organization's official number independently (not from the caller)

3. Call back and ask to verify the situation

Real institutions will understand. Scammers won't survive this simple check.

Limit Public Audio

AI voice cloning trains on available audio. Consider:

• Limiting video posts on social media, or keeping them private

• Being cautious about voicemails that could be recorded

• Understanding that podcasts, interviews, and public speaking create training data

Talk to Aging Parents

Many phone scams specifically target seniors, who may be less familiar with AI capabilities. Have honest, non-condescending conversations about:

• The existence of voice cloning technology

• The family code word system

• The permission to always verify before sending money

Report Everything

If you receive a scam call, report it to:

• The FTC at reportfraud.ftc.gov

• The FBI's IC3 at ic3.gov

• Your phone carrier

Reporting builds the databases that help everyone—and helps law enforcement identify patterns that lead to prosecutions.

Frequently Asked Questions

Can AI really clone someone's voice from just a few seconds of audio?

Yes. Current AI voice cloning technology can create convincing replicas from as little as 3-30 seconds of source audio. A 2024 report in Fortune noted that voice cloning has crossed the "indistinguishable threshold" for ordinary listeners.

Why don't phone carriers do more to stop scam calls?

Carriers face technical and regulatory challenges. While STIR/SHAKEN technology helps authenticate call origins, implementation remains incomplete. According to U.S. PIRG, only 44% of phone companies have fully deployed mandated anti-robocall software as of 2025.

Are younger people also at risk, or is this mainly a problem for seniors?

Interestingly, Truecaller's research found that young adults (ages 18-44) are actually three times more likely than older adults (45+) to have lost money to a phone scam. However, when seniors are victimized, the financial impact is typically much more severe.

What's the difference between robocalls and vishing?

Robocalls are automated, pre-recorded messages. Vishing (voice phishing) involves real-time human or AI interaction designed to manipulate the victim through conversation. Vishing attacks have grown 442% recently because they're more effective—and AI is making them scalable.

Can my bank help if I'm scammed?

Recovery rates for sophisticated phone scams are below 5%. Wire transfers and cryptocurrency payments are particularly difficult to reverse. Prevention is far more effective than attempting recovery after the fact.

Why We Started Lemma

We built Lemma because we believe the phone line is the last unprotected door—and it's time someone closed it.

The problem isn't that people are careless. It's that they're facing a new generation of threats with no real protection. When one in three unknown calls is spam or fraud, and when AI can clone your daughter's voice from a Facebook video, the old advice to "just be careful" isn't enough.

We're building AI-native phone protection designed from the ground up for this new reality:

• Intelligent call screening that goes beyond number databases to analyze patterns and behavior

• Real-time conversation monitoring that detects manipulation tactics, artificial urgency, and known scam patterns

• Transparency that shows you how decisions are made and explains what threats were blocked

• Protection for both directions—not just the calls coming in, but the calls you make

The technology to make phone calls safe exists. What's been missing is focus.

Everyone is thinking about cybersecurity on the internet. Lemma is thinking about cybersecurity on the phone line.

Because every unknown call shouldn't be a gamble. And your phone should be a tool for connection—not a source of anxiety.

If you're interested in protecting your family or business from phone-based threats, join our early access list or explore how Lemma works. The back door doesn't have to stay open.