3 ms·

I doubt this story is true. That’s not to say this specific scam doesn’t happen, it does, this exact scam — stuck in jail, stuck in a foreign country — is rife

I doubt this story is true. That’s not to say this specific scam doesn’t happen, it does, this exact scam — stuck in jail, stuck in a foreign country — is rife across all mediums, but the sophistication required to introduce artificial is not necessary. 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).

rom-antics 4y ago
> we are not yet at the point where a voice can be mimicked accurately from a voicemail or a clip on social media See: https://www.thestar.com/business/technology/2023/01/10/microsofts-new-ai-can-mimic-any-voice-after-just-3-seconds-and-some-experts-are-worried.html https://www.thestar.com/business/technology/2023/01/10/micro... > Microsoft’s new AI can mimic any voice after just 3 seconds Play the result over a low-bandwidth PTSN line and it could probably fool people in tech, let alone your grandma. Welcome to the future.
- phphphphp 4y ago
  That’s a paper with very carefully selected examples. The point I’m making is not that this will never be possible (it will) but rather it is not happening right now and this article is just victims and journalists doing exactly what all scam victims do: look for a reason they fell for the scam that softens the blow. AI in the news? Must be AI! Scams don’t need sophisticated technology to work, suggesting that scammers have access to some state-of-the-art technology that has only been demonstrated in a paper doesn’t match up with the reality of what scamming is. Getting scammed is embarrassing and our natural response is to look for reasons beyond the obvious. These scammers don’t need state of the art voice technology, because this exact scam has been working for the best part of a decade across many different mediums. There’s a reason if you walk into a shop and try to buy $1000 in gift cards, or walk into a bank and try to withdraw $3000 in cash, that the teller will assume you’re the victim of a scam! Protecting ourselves and our families from being scammed requires a sensible, realistic understanding of how they work: getting caught up in some AI fantasy misses the forest for the trees. Go back to the month before ChatGPT was unveiled and you’ll find this exact same news article written weekly for the last 5 years, except it won’t mention AI, because it wasn’t hot then. https://consumer.ftc.gov/articles/scammers-use-fake-emergencies-steal-your-money https://consumer.ftc.gov/articles/scammers-use-fake-emergenc... https://www.aarp.org/money/scams-fraud/info-07-2010/scam_alert_stranded_in_london.html https://www.aarp.org/money/scams-fraud/info-07-2010/scam_ale... https://theguardian.com/money/2013/nov/13/stranded-traveller-phishing-scam https://theguardian.com/money/2013/nov/13/stranded-traveller... https://travel.state.gov/content/travel/en/international-travel/emergencies/international-financial-scams.html https://travel.state.gov/content/travel/en/international-tra...
flangola7 4y ago
>but we are not yet at the point where a voice can be mimicked accurately from a voicemail or a clip on social media. Have you tried using Eleven AI? We are very much at this point.

rom-antics 4y ago

> we are not yet at the point where a voice can be mimicked accurately from a voicemail or a clip on social media See: https://www.thestar.com/business/technology/2023/01/10/microsofts-new-ai-can-mimic-any-voice-after-just-3-seconds-and-some-experts-are-worried.html https://www.thestar.com/business/technology/2023/01/10/micro... > Microsoft’s new AI can mimic any voice after just 3 seconds Play the result over a low-bandwidth PTSN line and it could probably fool people in tech, let alone your grandma. Welcome to the future.

phphphphp 4y ago
That’s a paper with very carefully selected examples. The point I’m making is not that this will never be possible (it will) but rather it is not happening right now and this article is just victims and journalists doing exactly what all scam victims do: look for a reason they fell for the scam that softens the blow. AI in the news? Must be AI! Scams don’t need sophisticated technology to work, suggesting that scammers have access to some state-of-the-art technology that has only been demonstrated in a paper doesn’t match up with the reality of what scamming is. Getting scammed is embarrassing and our natural response is to look for reasons beyond the obvious. These scammers don’t need state of the art voice technology, because this exact scam has been working for the best part of a decade across many different mediums. There’s a reason if you walk into a shop and try to buy $1000 in gift cards, or walk into a bank and try to withdraw $3000 in cash, that the teller will assume you’re the victim of a scam! Protecting ourselves and our families from being scammed requires a sensible, realistic understanding of how they work: getting caught up in some AI fantasy misses the forest for the trees. Go back to the month before ChatGPT was unveiled and you’ll find this exact same news article written weekly for the last 5 years, except it won’t mention AI, because it wasn’t hot then. https://consumer.ftc.gov/articles/scammers-use-fake-emergencies-steal-your-money https://consumer.ftc.gov/articles/scammers-use-fake-emergenc... https://www.aarp.org/money/scams-fraud/info-07-2010/scam_alert_stranded_in_london.html https://www.aarp.org/money/scams-fraud/info-07-2010/scam_ale... https://theguardian.com/money/2013/nov/13/stranded-traveller-phishing-scam https://theguardian.com/money/2013/nov/13/stranded-traveller... https://travel.state.gov/content/travel/en/international-travel/emergencies/international-financial-scams.html https://travel.state.gov/content/travel/en/international-tra...

flangola7 4y ago

>but we are not yet at the point where a voice can be mimicked accurately from a voicemail or a clip on social media. Have you tried using Eleven AI? We are very much at this point.