3 AI scams (plus one extra creepy AI app) you should be aware of

On this blog, I try to keep things relatively upbeat.

But we are all grown-ups, so it’s time to discuss some of AI's downsides because that stuff is already out there and already happening.

It mostly concerns AI-generated voice and video scams that mimic the voices of people close to the target, such as family members or friends.

Because AI is getting so good right now, it’s often hard to tell the difference between what is fake and what is real; as you’ve seen on this blog, a single 30-second voice recording can be enough for an AI to clone a voice convincingly.

The video below is doing the rounds at the moment, it says five AI Scams on the tin; but really there are three that I’ll summarise below:

1. Voice cloning

Someone cloned tech investor Martin Casado’s voice using Generative AI and tried to scam his father (luckily, it didn’t work!).

This takes the whole “hey mum” text scam to a new level - as it is getting easier to recognise when a text is fake. But this is a whole new level which includes a voice that sounds real. Read more about that one here.

2. Video cloning

A Hong Kong multinational firm finance worker was tricked by fraudsters who used deepfake video technology.

Hong Kong police told CNN the scam involved the worker attending multiple Zoom calls with what he believed were other staff members but were actually deep fake re-creations.

This almost cost the firm $25.6m USD.

Could real estate be targeted similarly?

3. Emotional attachment

Finally, the video suggests that we might all become emotionally attached to the bots we work with, who might then be able to scam us.

Before you think about how far-fetched that sounds, have you seen the movie Her, starring Joaquin Phoenix?

Joaquin’s character, Theodore, ends up in a relationship with a highly emotionally intelligent chatbot named Samantha (I’m not kidding about the relationship or her name!).

Note: This isn’t the first time it’s been suggested that we may end up liking our AI friends more than our human ones - just because they seem to get us more.

Interestingly, this article suggests that a ‘Samantha’ could be built to run a real estate transaction. I’m not convinced—yet.

And now for the creepy app…

This is a bit freaky. Take a photo of a place and upload it to Geospy.ai. It will try to work out where on Earth the photo was taken.

Here’s my test; it’s pretty accurate, so I guess there goes just a little bit more of whatever privacy we might have had left.

So - what can we do to protect ourselves?

First up is awareness and education

Understanding the capabilities and limitations of AI is the most important thing.

I’ve said this before, but we should educate ourselves and our kids about the nature of interactions with tools like ChatGPT, Claude, Hey Pi, etc., and the fact that they are not genuine human connections.

Supervision/Monitoring

Whether in business or at home, I would monitor what tools your kids and/or employees use and openly discuss them. How are they useful? Are they saving time? Is your data safe?

Critical thinking skills

Encouraging healthy scepticism about information and interactions can help users question the authenticity and motives behind communications that seem a little suss. We’ve learned to sort the fake Nigerian princes in emails from the real ones; now it’s time to do the same with AI.

Finally - always think before you upload your geolocation to social media.

Happy Hunting 🚀