For millions of people around the world, artificial intelligence is making life easier in countless ways. It can summarize emails, generate images in seconds, help students study, and even make streaming services better at recommending what to watch next. Unfortunately, the same technology is also making life easier for scammers.
A new Insuranceopedia study puts a startling price tag on one of AI’s fastest-growing threats, estimating that deepfake voice scams could cost Americans nearly $40 billion a year if current trends continue. That’s an eye-popping estimate, but it’s based on a growing body of research showing that AI voice cloning has become remarkably convincing and increasingly common.
Insuranceopedia says 25% of Americans received a deepfake voice call in the past year, which translates to roughly 64.5 million adults being targeted. The study also notes that 77% of people who received an AI voice scam reported losing money, and it uses the FTC’s median reported imposter-scam loss of $800 to arrive at a base-case estimate of $39.7 billion. Even the conservative scenario still lands at $2.6 billion in losses, while the high case climbs to $46.4 billion.
Insuranceopedia based its analysis on several recent studies, including research from Hiya, McAfee, the FTC, and U.S. Census data. To put that figure into perspective, Insuranceopedia found that $39.7 billion would be enough to:
- Pay the average annual salaries of roughly 750,000 Americans.
- Buy nearly 880,000 new vehicles priced at $45,000.
- Cover the median value of more than 95,000 U.S. homes.
Those comparisons drive home just how much money could be at stake if consumers aren’t prepared.
AI is changing the internet faster than most people realize
Those sceneario estimates show how quickly quickly costs can escalate as AI becomes more accessible. The simple reason why the deepfake voice scams are so effective is because they do not sound like obvious fraud anymore. In some cases, scammers can use AI to imitate a family member, a boss, or a public figure well enough to trigger an emotional response before the victim has time to think. The FTC has warned for years that voice cloning can make family-emergency scams more convincing, and that a short audio clip can be enough for scammers to create a believable fake.
In today’s tech-centric world, deepfakes are already showing up in fake videos of politicians, journalists, and public figures, and YouTube has been expanding its enforcement and likeness-detection tools to help remove deceptive synthetic media.
And while AI can make scams cheaper to run, it also makes the infrastructure around them more expensive to power. A previous study found that AI’s growing energy appetite is putting more pressure on electricity demand and, by extension, household bills. So the technology boom is creating costs on both the consumer side and the scammer side of the equation.
The good news is that the fight back has already started
Artificial intelligence is also reshaping industries in ways that go far beyond scams as companies and agencies that oversee digital communication are beginning to respond more aggressively.
As previously noted, YouTube has rolled out an AI likeness detection tool that lets creators search for content where their face has been used in an AI-generated or altered way. The company says the feature is meant to help users protect their image while still allowing productive This tool gives people a way to respond after a deepfake spreads, instead of leaving them to chase it after the damage is done.
The FCC is also trying to tighten the screws on scam infrastructure. In April, the FCC adopted new rules to strengthen Know-Your-Customer requirements for voice service providers, saying it wants to stop illegal calls before they enter the network and give consumers more information about who is calling. Then in May, the agency proposed more specific rules to hold voice service providers accountable for illegal robocalls and to strengthen the STIR/SHAKEN framework used to verify caller ID.
These changes are important because the deepfake problem and the robocall problem are converging. Scammers are not just spoofing numbers anymore. They are spoofing people. The more the system allows bad actors to hide behind fake identities, the easier it becomes to blend AI voice cloning with the older scams consumers already know too well.
How to protect yourself from AI scams calls
The best defense is still a mix of skepticism and simple habits. As mentioned before, the FTC says a scammer can clone a loved one’s voice from a short audio clip, and the agency’s straightforward advice is do not trust the voice alone. If a call sounds urgent, emotional, or unusually convincing, hang up and verify it using a number you already know is real.
A few other steps can make a big difference:
- Call back using a known number. Do not rely on the number that called you. Scammers can spoof caller ID.
- Create a family safe word. If someone claims to be a relative in trouble, ask for the word before taking action.
- Never rush into a payment. Gift cards, wire transfers, and cryptocurrency are major red flags in scam situations. The FTC warns that scammers often push people to move money quickly.
- Verify emotional requests in a second channel. If a message is scary or unusual, confirm it by text, another phone number, or another family member before sending money or information.
- Report the scam. The FTC encourages consumers to file complaints at ReportFraud.ftc.gov, so investigators can track patterns and spot emerging fraud.
That kind of caution may sound basic, but basic is exactly what works when the scam is built around panic and speed.
AI isn’t slowing down and neither are the scammers
The Insuranceopedia study isn’t a prediction of what Americans will lose this year. It’s a reminder of how quickly AI-powered scams are evolving and why consumers need to evolve with them. As AI becomes a bigger part of everyday life, learning how to recognize these scams may become just as important as learning how to use the technology itself.
Until the guardrails catch up, the Insuranceopedia study is a reminder that AI can also help criminals move faster, sound more convincing, and reach more victims. The numbers show just how expensive that shift could become if consumers are not careful and regulators do not keep pace. The ability to pause, verify, and think twice before reacting may become one of the most valuable digital skills anyone can have.