A Familiar Voice Is No Longer Proof

For as long as there have been people, hearing someone's voice was proof of who they were. It was the oldest identity check we had, it required no technology, and it worked. That ended quietly, in about two years, and almost no family has changed a single habit in response.

The numbers are worse than most people realize. Americans 60 and older reported $7.7 billion in losses to the FBI's Internet Crime Complaint Center in 2025, roughly 60% higher than 2024 and more than double the $3.4 billion reported in 2023. The average loss for an older victim now exceeds $38,000, close to twice what younger victims lose, and more than 12,400 people over 60 lost over $100,000 each. Of the $893 million the FBI ties specifically to AI-enabled scams, $352 million came from people over 60. The American Bar Association notes that cloning a voice convincingly takes about three seconds of audio, which social media hands over for free.

Here is why the usual advice has quietly become useless. Nearly every warning ever written about scams is a detection strategy. Look for the odd phrasing. Notice the urgency. Trust your gut. All of that assumed the fake would be imperfect. It no longer is. You cannot out-notice a flawless copy of your granddaughter crying at two in the morning, and it is not a failure of intelligence or attention when someone believes it. That is the entire product.

What still works is structural instead of vigilant, and DailyCaring ran a useful checklist of it this month. Agree on a family code word that any emergency money request has to include. Adopt a hang up and call back rule, always dialing a number you looked up yourself rather than one you were given. Name a trusted contact at the bank or brokerage, which most institutions allow and which takes about five minutes. Practice on real examples together. (Worth knowing: that article's author runs a phishing-practice company, so weigh the recommendation to train weekly on a platform accordingly. The code word, the callback, and the trusted contact are standard, free, and the part that matters.)

The reason those work is that they do not depend on being right in the moment. A code word holds whether or not the voice fooled you. That is the whole design principle. Stop trying to detect, and start building something that survives being deceived.

Which brings up the part nobody writes down. Setting any of this up requires a conversation that quietly says, you could be fooled. To someone already losing pieces of their independence, that lands as one more vote of no confidence, and it is exactly why these conversations get postponed until after the money is gone.

So do not frame it as protection for them. Frame it as a family policy, because it honestly should be. We are picking a code word. It applies to everyone, including me, and especially to me, because I am the one who travels and might actually be the person calling in a panic someday. Then go first. Give them your code word before you ask for theirs.

The technology that broke this took two years. The conversation takes about ten minutes.