The Family Safe Word: A Simple Defense Against AI Voice Scams

Every documented AI voice cloning case follows the same shape: a familiar voice, a crisis, and pressure to act before you can think. Consumer protection groups and law enforcement now recommend the same low-tech countermeasure against a very high-tech problem - a family safe word.

Why you can't just "listen closer"

It's tempting to think you'd notice if it wasn't really your child, your parent, or your grandchild on the phone. Surveys on the topic keep landing in the same place: most people say they aren't confident they could tell a cloned voice from the real thing. The families in documented cases like Jennifer DeStefano's and Marilyn Crawford's didn't hesitate for a second, because the voice was, in every detail that mattered to their ear, correct. Modern cloning tools only need a few seconds of audio - often lifted from a public social media video - to reproduce pitch, cadence, and the small verbal habits a parent would recognize instantly.

You cannot train your ear to beat this. You can build a habit that doesn't rely on your ear at all.

The safe word protocol

• Pick a word or phrase with your family now - not during a crisis. Something unrelated to names, birthdays, or anything easily guessed or found online.
• Never share it over text, social media, or anywhere it could leak. Say it out loud, once, in person.
• If anyone calls in distress asking for money, ask for the word before doing anything else. A real loved one will know it instantly. A cloned voice - however perfect - won't.
• If they can't give it, hang up and call them back directly on a number you already have, not one the caller gives you.

Beyond the safe word

• Slow down on purpose. Every one of these scams is built around urgency. A real emergency can survive a five-minute callback.
• Don't confirm details for the caller. Scammers often already have basic information - a name, a city - and use it to sound credible. Repeating it back to them isn't the same as verifying it.
• Talk to the people most likely to be targeted, before they're targeted. Older parents and grandparents are disproportionately targeted, and the conversation is far easier to have calmly, in advance, than in the middle of a panicked call.
• Treat "don't call anyone else" as the biggest red flag of all. No legitimate emergency depends on you not verifying it.

None of this requires new technology - just five minutes with the people you'd want to protect. It's also exactly the kind of thing we think shouldn't be locked behind a product. Vocos is being built in the open so that better defenses than a safe word - ones that can listen to the call itself - are available to everyone, not just people who can afford them.