Dirtyhash.com: An open source project to defend against Web3 and Crypto frauds

[adroitanka]

I recently came across https://dirtyhash.com/, an open source project to defend against Web3 and Crypto frauds, phishing, ransomware etc. With this you can also analyze suspicious entities (wallet addresses, smart contracts, web3 and regular domains, NFTs, social media handles) using machine learning algorithm and get risk scores. I tried few bitcoin addresses and results look convincing. Moreover, they have a feature to report any scams and also share them with the community. Summary with some more details in an medium article: https://medium.com/@amiacs/towards-a-safer-web3-defending-against-cryptoscams-with-dirtyhash-b0d08f1039e6

What do you guys think? Can this be a useful tool to detect crypto fraud? 

[examplens]

Is this a tool of yours that you want to present or are you just curious?
even if it is your idea, there is no need to hide it, here you can get some useful guidelines for sure. in the article, I saw some impressive figures about the use of this service, although as far as I can see the website itself is very young, so it seems illogical to me.

What do you guys think? Can this be a useful tool to detect crypto fraud? 

so far I have not seen a single automated AI service that can completely replace human reasoning. that's why I wouldn't trust this service of yours either, but it can help maybe.
at some other ends of the chain, we fight with those who abuse results from similar services and keep someone's money under the pretext of dirty bitcoin.
For example

[adroitanka]

This is an idea of an acquaintance and so when he told me about it, I was wondering how reliable and useful such a service could be and completely agree AI can never replace human reasoning. I will let him know about this post here in the forum. May be people more experienced and into the field can give guidelines.

[examplens]

This is an idea of an acquaintance and so when he told me about it, I was wondering how reliable and useful such a service could be and completely agree AI can never replace human reasoning. I will let him know about this post here in the forum. May be people more experienced and into the field can give guidelines.

it would certainly be good if he came and explained the principle of the functioning of his algorithm and how exactly the results are obtained. I believe that he will get additional useful ideas here, after all, many heads are smarter than one.

I'll post the reference once more, maybe you haven't seen it. Where one exchanger (pay attention to user @fixedfloat) claims that some of their customers use funds that were obtained fraudulently. and that's why they keep the funds and send them to the address of the "victim" without explaining what method they use to reach their conclusion. sounds crazy, doesn't it?
check here https://bitcointalk.org/index.php?topic=5424907.0

[amiacs]

Hi, I am from DirtyHash team. Thanks for the discussion on this topic here.
The use of the service is real, and the information provided in the medium article is correct. Please reach out if you have aby specific questions.

For the DirtyHash risk scores - it is based on 1) other users reporting an address as fraud/scam (the more users report, the higher the risk score), and 2) machine learning. We have created a dataset of ETH and BTC fraud and non-fraud addresses, used that dataset to learn fraudulent behavior. The accuracy of the algorithm is pretty good. We will also write a post describing in detail the algo specifics.

I am open to any feedback that you might have, and happy to discuss with you.
DirtyHash project is open source, you can find it on Github.