Show HN: I built a tool to detect scams | Hacker News Reader

Show HN: I built a tool to detect scams

scamchecknow.com

30 pointsvinnythejinny31 comments

Nice tool, however if I am correct you are calling the Google generative language API directly, your API key is exposed to the client (the browser).

If you have no spending limits, this is a very (very very very) bad idea.

damn, I can't stress this enough. You publish a key be prepared for millions on dollars of issues.

Yes, exactly. Google seems to be calm in this regard luckily as far as i know though.

OP please revoke the key ASAP

Done. Thanks both of you for pointing out my obvious mistake

Thanks a lot man. I totally missed it lol. Just fixed it.

After a quick look: input seems to be a screenshot - but this would omit quite a lot of useful information.

E.g.

- email app might not show the actual sender email adress but just their "name"

- links can look legit on a screenshot (e.g. yourBank.com) but the actual linked address is scamYou.com.

Given these obvious limitations, I'm really doubtful how trustworthy this solution can be.

Plus:

- the contact page does not show any names, just claims a " team of cybersecurity experts "

- about page in the footer 404s out.

This could as well just be a scam itself to read others' emails.

Haven’t tested but I’d just like to commend anyone doing something to spoil the success rate of scammers who prey upon the old, the gullible, and the just plain too busy to check properly.

People who think they're too busy need to make time in proportion to the money involved. I read a news story about someone scammed out of the price of a home because they were on holiday and too busy to be suspicious that the realtor suddenly emailed to ask them to send the money to a different account.

Where am based (Singapore), last year in 2024 alone people lost over $1.1 Billion dollars. Some of the cases are so extreme where the bank is trying to tell the person that they are getting scammed but they wont even believe the bank officers. Manipulation is the key here

They also prey on the lonely. The big new thing in scamming is "pig butchering". Make friends with someone online, gradually persuade them you're a genius crypto trader, get them depositing funds to a "crypto trading platform", number goes up, get them depositing more funds... then pull the rug.

https://www.youtube.com/watch?v=1QhMqoTNSl8

I remember when I was young in the early 2000s, the internet was a place you could make friends. These pig butchering scammers are working to wreck what little social fabric the internet has left.

That's not a new thing at all. Most successful scams have been around for centuries in one form or another. They just get renamed/brought back up when new technology gets involved.

Guys, Thanks for the feedback so far, am glad most of you like the service. Thanks for those who found the bug with my api and its fixed now. Would love any other feedback regarding the features and if anything else you would like

The thing I worry about with this kind of thing is that without very careful UX integration, people take false negatives as proof that things are OK, rather than treating it as just one of many signals.

Also. Is this tool for sale so I can hide only my scams in it?

you mean a paid version or private just for you ?

Have you tested it ? I suggest you test it with multiple screenshots (literal scam text, just a normal conversation (not a scam) and borderline doubtful conversation. The AI is pretty good at identifying them, with a confidence score and risk level.

Won't the scammers not just use this to improve their campaigns?

This is exactly what I thought.

Another domain, but the same result [1]. Better detectors just yield better fake.

1: https://proceedings.neurips.cc/paper/2019/hash/3e9f0fc9b2f89...

In some cases there's an inherent tradeoff.

A scam might, for instance, say "If you don't go to phishingpage.xyz RIGHT NOW your account will be DELETED!" to capitalize on urgency. That appeal to urgency can be detected. The scammers could of course just choose not to appeal to urgency, but then that would reduce the number of people who'd fall for their scam.

Kindly... (SCAM!)

Dear sir... (SCAM!)

Voicemail half cut off starts talking about "I see your papers on my desk here" (SCAM!)

It is amazing though Twilio, you can write a voice mail via XML and start dialing away.

lol maybe i should do that next. Voice based scam identification in real time is the need of the hour

Built on Lovable? Looks neat. If yes, how was your experience? Did you have to customize much by hand or did it take care of most of the work?

It was great, am glad it saved a lot of my time. Of course prompts are important but it would surely cut down your overall dev time.

Nice, tried with a scam text I received today. 95% sure it was scam.

It gives itself a 60% Safe rating.

Is there an API that i can use?

An API would be awesome.

Would look into it. Thanks for the idea.

scamchecknow[.]com Creation Date: 2025-03-07T11:59:10Z

Looks like red flag.

Down?

working fine at my end.

Reply on news.ycombinator.com