Yesterday, the identity verification company Veriff dropped a report that should be sitting on every dinner table in America. Across a study of 3,000 adults run with Kantar, U.S. respondents scored a 0.07 on deepfake detection, on a scale where zero means pure guessing. When the test showed a pair of videos with one real woman and one AI fake, seventy percent of Americans pointed at the fake and called it real.
That is not a margin. That is a coin flip with the worse side up. And it landed in the middle of the AI deepfake risk conversation at exactly the moment most people still believe they would know a fake if they saw one.
Half of U.S. respondents told Veriff they were confident in their ability to spot manipulated media. Their actual performance came in near random. The gap between what we think we can see and what we can actually see has become its own threat surface.
This is not the first report to flag the issue, but it might be the cleanest snapshot we have of where the public sits in 2026. Veriff ran the same quiz with adults in the United States, the United Kingdom, and Brazil. Americans came in last on awareness of the term "deepfake" itself, at 63 percent familiar, trailing British and Brazilian respondents by a meaningful margin.
Why videos? Because the cost of producing them has collapsed. What used to take a studio now takes a free app and a clip from social media. The financial loss data tells you what that collapse has produced. Deepfake-driven fraud has caused more than $2 billion in global losses, with the United States taking the largest national hit at $712 million.
Here is the part that should worry community leaders, parents, executives, and anyone over 60 with adult children. The Veriff study found Americans are also more likely than respondents in the UK or Brazil to trust social platforms and digital services to flag AI-generated content on their behalf. That trust matters because it lowers individual vigilance. If you believe the platform has your back, you click faster, share faster, and verify less.
The FBI has been pointing at the same wall for months. In a recent public service announcement, the Bureau described an active campaign of malicious actors impersonating senior U.S. officials using AI-generated voice messages, sent to federal and state officials and people in their contact networks. Voice cloning now requires only a few seconds of audio. A short voicemail. A clip from a podcast. A reel from a niece's birthday party.
The Bureau's 2025 Internet Crime Report logged more than 22,000 AI-related fraud complaints with losses exceeding $893 million. The fastest growing wedge inside that figure includes scams that use a cloned voice to fake a kidnapped grandchild or an executive demanding an urgent wire transfer.
Pull the headline numbers off the page and the picture turns very human very fast. A mother gets a call from a daughter's number, hears her voice in distress, and is asked to wire money. A CFO gets a video call from a CEO who sounds and looks right, and approves a transaction that drains a quarter million dollars out of an operating account. A teenager finds a fabricated explicit clip with her face on it circulating through a school group chat.
These are not edge cases. They are the new shape of AI deepfake risk, and they live inside the same gap the Veriff report just measured.
State legislatures are moving. Maryland became the 30th U.S. state to enact election deepfake legislation, with its measure taking effect June 1. Washington and Pennsylvania both activated new deepfake liability frameworks this spring. The statutes vary, but the throughline is consistent. Courts now have language to assign liability when someone's face, voice, or likeness is used without consent.
Federal action is moving slower. Insurance carriers are watching but largely on the sidelines. Which is why community matters now. People are walking into AI deepfake risk faster than the rules can be written.
We are building toward solutions. That includes working with policy folks, technology partners, and the right insurance partners to shape what coverage and recovery should look like in a world where a stranger can clone your voice from a YouTube comment. We are not selling a policy. We are building the community that is going to lead this conversation, and the InsureMyAvatar community is where it happens.
For now, the most useful thing you can do is the simplest. Set a family safe word. Treat any urgent money request as automatically suspect until you call back on a number you already had. Talk with the older adults in your life about voice cloning, and with the younger ones about consent and image misuse. Pay attention to which state laws apply to you.
The coin flip in the Veriff report is not a verdict. It is a starting point. Once people see how thin the line between real and fake has become, they start asking the right questions. That is where every form of protection has always begun. Awareness is itself a form of armor, especially in a year where most Americans cannot tell the difference between a real face and a generated one.
Take our free risk assessment at insuremyavatar.com and sign up for weekly digital safety updates, news, and tips you need to know.