Inspiration
Many times as a kid, I would want to get games for free like Geometry Dash and Minecraft. In order to do this, I tried to pirate them from sketchy websites. I also used to believe scammers in emails sent to me about winning a lottery or prize. Every time my computer broke and scammers won, I would get really angry. That's why I built PrivacyGuard. Whenever I see a website or an email that I am unsure about, I could just check PrivacyGuard and know whether it's safe in less than a second.
What it does
PrivacyGuard has 6 features:
- Sensitive Information Detector
- Cookie Deleter
- Live Privacy Dashboard
- Camera Detector
- Site Safety Rater
- Scam Message Detector
How we built it
I first started off by building a github repository, then started walking around the room and gathering ideas for what people would actually like to see to improve their safety. I also came up with some ideas of my own. Then, for each idea, I asked an AI to generate a template on how I could actually build the feature, and then had the AI walk me through each step of creating the features. Then after each 5 or so lines of code, I would add a comment explaining what it does. Eventually, I ended up with 6 features that all go into deep depth while still being accurate, efficient, and fast.
Challenges we ran into
The main challenge I ran into was detecting whether a site was sketchy, because I wanted to preserve the safety of the user while also being able to thoroughly give proper advice on the website. At first, it gave every site 100 because the only thing it graded on is whether it is hosted on HTTP/HTTPS or IPv4/6. I then had to fix the feature for over an hour, in an attempt to not only judge the link itsle,f but the site content. After a while, the site as able to detect malicious downloads, bad links, and more, and give a proper rating.
Accomplishments that we're proud of
I'm really proud of the scammer detection system. I got over 4000 words and key phrases that indicate a scam email or text message in less than a second. This was super helpful to anyone, because a scammer always makes the matter seem urgent, so having a long slow process would by then make the victim already pay. I'm really proud of this because I think that this fast process will stop people for falling for scams due to time pressure.
What we learned
I learned a lot of new things. First, I learned how to use GitHub to a level I never knew existed. I learned how to add screen, format HTML and CSS, write some javascript, and create an overlap so that they work together hand in hand.
What's next for PrivacyGuard
I will plan on adding a layer of neural ntworks to the scamd etection system so that the scam detection improves with the scammers adapting. The neural netowrk will be trained to become really strong at identifying even the smallest sign of a scam to save the victim from losing money.
