I found weird #AI #slop in my search results again. This time, it's a website called scam-detector[dot]com. Scam-detector is a service that lets you find out if a website is a scam - all you have to do is provide a URL. Sources basic data like domain age and public block lists as well as — and this seems to matter way more — asking an LLM. It then computes a score from 0(?) to 100 and fills in a template.
I fed it some stuff from #Phishtank. The first site it didn't like because it saw a login form (clearly pointing at adult content). I then gave it a site (from Phishtank again) that promised to pay you for watching YouTube. The LLM was not critical at all, although the domain age made the score terrible.
Up next, I grabbed a Microsoft phish from @urldna. The result? A score of 80 points!
What's slightly more crazy is that the website was plastered with green button PNGs which were ads for #Guardio, #Incogni and seemingly themselves. “Are You A Crypto Fraud Victim? CLICK HERE” it said. Naturally, I did and was taken to a form that offered me to recover my crypto. All it needed was my first and last name, my email, what I invested, my country and city, my countrie's phone code, my phone number (yes, seperately), links to the website as well as a contact person to the scam website.
@Mensh123
Now put that ad into the scam detector!
@libewa I did that and it scored it 100 (perfectly legit). The funny thing is that this might not even be hardcoded, the AI just seems to be extremely trusting when it comes to anything a website says and the domain is neither extremely fresh nor on major blocklists. My curiosity did however lead me to discover that the site has a comment section where it itself is critisized by a significant number of commentors.