Exactly four years ago, Mark Zuckerberg announced a revolutionary product, Ray-Ban Wayfarers with two cameras and a microphone inside. At the time, nobody cared because the main use case was to live stream your boring life on Instagram Live, but in 2024, these smart glasses are back and loaded with game-changing AI features. Like when you wear them, your brain's memory becomes obsolete. When you lose your car in a parking garage, there's no need to panic because your glasses remember everything. Hey Meta, remind me where I parked my car.
You parked in spot 2304. Oh, okay. It's kind of weird how as AI gets better, Zuckerborg appears more human. But the glasses can also do things like language translation, which means learning a new language is a total waste of time now. Most importantly though, the glasses help you understand the world around you.
Like, what type of animal is this? That's a horse. Pretty useful and innocuous, but what if you could look some random stranger in the eyes and figure out their name, where they work, their social credit score, what they ate for lunch, and all sorts of other data they thought was private? In what feels like multiple Black Mirror episodes coming to life, someone actually built an app for that. In today's video, we'll try to reverse-engineer it, and you'll learn how to protect your own data to avoid becoming a part of this rapidly approaching dystopian future.
It is October 7th, 2024, and you're watching The Code Report. Google was the first to market with smart glasses over a decade ago, but they made a critical mistake when designing the Google Glass product. The end user looks like walking cringe when wearing them.
Like all meta products and features, Zuck stole the idea and then removed the clown energy from it to make it a viable product. In this case, they partnered with Ray-Ban to make the glasses just look normal. Now, smart people don't wear smart glasses, but I bought my pair for $299, and they're absolutely amazing for violating the privacy of random NPCs in my environment.
I can now easily walk up to a random stranger and start a conversation, because just by looking at their face, I know their name, where they went to school, where they work, their recent DUI arrest, and all sorts of other great conversation starters. So here's how it works. We stream the video from the glasses straight to Instagram and have a computer program monitor the stream. We use AI to detect when we're looking at someone's face. Then we scour the internet to find more pictures of that person.
Finally, we use data sources like online articles and voter registration databases to figure out their name, phone number, home address, and relatives'names. This is made possible thanks to an app called iXray. which works automatically on the fly by combining facial recognition software with LLMs and public databases. Luckily, though, the app's not built for public consumption, but rather to highlight the insane privacy concerns this type of technology presents. The details are sparse, but they describe how a pipeline of five technologies were used to build it.
They don't disclose these technologies directly, because they know I would just make my own tutorial about how to build your own. but we can make an educated guess about what they might be. Here's how I would build it.
The glasses provide a real-time video feed, and the first step is to detect a face. There are numerous open-source vision models to accomplish this that return you with coordinates for all the facial features in an image, but these models won't tell you who the face belongs to. For that, we'll need to move on to step two, a reverse image search using a tool like PIMEyes or FaceCheckID. These technologies are highly controversial because they often contain images of people without their knowledge or consent.
and they're loved by stalkers and creeps who identify people from anonymous photos. If you upload images of yourself to the internet, you're likely on there. In any case, once we put a name to a face, we can move on to step three, using other controversial tools like Fast People Search or Instant Checkmate to give us information we shouldn't have, like the date of birth, phone numbers, addresses, and relatives of the person we're looking at.
At this point, we know we're looking at John Smith, age 23, from Booger Hole, West Virginia, and now we can move on to step four to perform a wider search on the web. using web scrapers or APIs to find matching LinkedIn profiles, Twitter accounts, and other socials that will give us a comprehensive understanding of our new best friend. It's a lot of data, so finally that brings us to step five, where we use a large language model like Llama to analyze all that data in seconds and return my phone with a not-creepy icebreaker.
Like, hey John, I'm so sorry to hear your dad passed away from lung cancer. I used to work with him in the coal mines, at which point we cry and hug and become best friends. It might be a little dishonest, but triggering a visceral emotional response is a great way to connect with people.
But at this point, you might be wondering, how do I wake up from this nightmare and protect my data? Well, the developers of iXray provide a guide that describes how to opt your data out of all these various data brokers. In most places, you have the right to be forgotten, and they need to delete this data when requested. But most people just willingly give up all their data to companies like Meta. And when you wear his glasses, Zuck has direct access to your balls.
The more your eyeballs look at Meta products, the more advertisements they can sell. But not everyone follows the law, and some entities are above the law. Like in China, they use technologies just like this to implement a social credit system. If you're worried about that in your country, Make sure not to share any photos of yourself online. Don't go out in public where CCT cameras might be able to capture you.
Don't get a government ID. Like this video, but then turn it off and get off the internet forever. And definitely don't wear the glasses.
This has been The Code Report. Thanks for watching, and I will see you in the next one.