They thought loved ones were calling for help. It was an AI scam

A little while later, I got another call from the "lawyer", with the correct caller ID. I played with that guy for a little bit too, and when he realized it he unleashed a string of competent profanity at me. When I responded calmly (I grew up on the Internet too, buddy), he broke character, told me he made tens of thousands of dollars per week, and offered me a job. I didn't really need any more excitement in my life so I politely declined to find out more and ended it, but it was certainly odd. Definitely a different class of scammer than the usual "medicare" and "buy your house" crap I get.

My plan is to create a set of code events from our past and to convince her to never trust “me” wrt finances, to always hang up and call my number and even then to leave an “incident” unresolved for at least one day.

I really didn't think we'd see it play out this fast, though.

Stop posting paywalled content. Your posts are just free advertising for a paid service.

Scams evolve out of necessity, the widespread nature and success of this scam without the use of artificial voices demonstrates that the scam does not need artificial voices.

The article mentions that in one of these cases, there’s very little audio available online of one of the victims and so it posits that even just a few sentences are enough to generate a real sounding voice… but doesn’t consider that maybe the voice was a generic man’s voice that, in a panic, their parents let their guard down and assumed that based on what the voice said that it must be their son.

There’s absolutely the technology out there to generate a real sounding mimic of someone’s voice given enough material, but we are not yet at the point where a voice can be mimicked accurately from a voicemail or a clip on social media.

The article is an example of overfitting: the last few months have been awash with news about technologies like Chat GPT but these voice mimicking technologies have been around for years now!

A far more likely explanation is that the scammers in these cases just happened to sound like the victims — after all, if you know someone’s gender, age and location, and set up a contrived situation in which audio quality is bad, you can guess what they sound like and explain away any discrepancy quite easily.

I guess my point is not that this isn’t going to happen in future, but that these scams don’t need something so sophisticated. People will fall for this scam without the need for AI (as they have been, for years).