AI deepfake scams: no hacking required, just your face and your voice

By Zendoric · October 1, 2026.

The US network 12News, based in Phoenix (Arizona), has devoted one of its consumer segments ("Ask the Experts") to explaining how scammers are turning to AI-generated videos (deepfakes) to impersonate other people. The original material is a local television piece with no accompanying written article containing verifiable data: there are no figures, specific cases or names we can attribute with any confidence. We therefore limit ourselves to what the headline and description allow us to state solidly: there exists, and it is news in the consumer slot of a regional television station, a deepfake fraud phenomenon widespread enough to warrant practical advice aimed at a general audience.

As industry context, this type of scam is not new in 2026. For several years now there have been documented cases of voice cloning to impersonate relatives in trouble (the so-called "grandparent scam"), fake videos of executives requesting urgent transfers ("CEO fraud") and ads using cloned celebrity faces to sell nonexistent products or cryptocurrencies. What has changed is not the fraud itself, but its production cost: generating a convincing cloned voice or a lip-synced video no longer requires technical expertise or a budget, just a few seconds of the victim's real audio or image, often taken from their own social media.

Our reading is that a piece like this -a consumer advice segment on local television, not an investigation by a major outlet- is in itself a useful indicator. When generative AI fraud reaches the "tips for the general audience" section of a regional newscast, it is a sign that it has stopped being a niche risk and become a problem of mass digital literacy, comparable to what email phishing was a decade ago. It is also exactly the kind of short-term cost that should not be downplayed within the transition toward more capable AI: the same technology that makes it possible to clone a voice to extort a family is the one that, properly governed, will allow films to be dubbed into any language or give a voice back to someone who has lost it to illness.

The underlying answer is not to halt the generation of synthetic content -unfeasible and, moreover, undesirable, given all the good that same capability also produces- but to build the verification infrastructure that is missing today: authentication of calls and videos, cryptographic provenance marks, and the still uncommon habit of confirming through a second channel before moving money. Until that infrastructure arrives, the usual advice against this type of fraud -be wary of any urgent request for money, however unmistakable the face or voice on the other end may seem- remains, for now, the most reliable defense we have.