Malware website filtering

lagomorph

As a comparison, QWANT has exteremly similar search rankings but has stripped out all the scam links.

Edit, this is no longer true - QWANT now also actively pushes these scam sites to top rankings.

yokoffing

Yes, we should have a way to report malicious links.

lagomorph I don't necessarily think that scam links should be fully purged from listings

I think they should be removed entirely from listing. A site that is obviously a scam or known to lead directly to viruses, malware, phishing, etc. should be removed.

Vlad

lagomorph

This is tricky to execute on.

Suppose we had a "flag as malware" system.. Then that means now we have to have dedicated people reviewing this all the time. We can not just automatically de-index something because someone flagged it as malware.

And suddenly it becomes non-scalalable, manual, operation that nobody want do so as the rules are not exactly defined and there is a lot of gray area what defines a malicious website.

To someone a major news site is malicious because all the tracking they do, and for someone else because they do not agree with its politics.

It is a very complicated problem.

You can imagine people flagging sites of their competitors.

You can imagine people flagging random websites they don't like.

And if we chose to ignore that, then why you have a system in the first place?

9rqnq

Vlad Does Kagi employ any mechanism to delist from index or block known phishing or malware websites? There are many threats that are relatively easy to be avoided with automatic filters. For example NextDNS use many filters to block such sites, like Google Safe Browsing, IDN Homograph Attacks, or block newly registered domains. All this settings are also optional. They also have their own AI model to filter suspicious sites, and I believe something like this should be possible to create in Kagi.

Personally, security is much more important to me than privacy. And I see privacy as a subset of security. Someone who doesn't have my personal data will have a much tougher time attacking me (through hacking or scams). But it is not impossible. Search engines are our window to information on the Internet. That's why they should not only be private but also secure. Elimination of advertisements is, of course, an important step towards more security, but it is not only necessary. And right now, I see only one SE that is capable of implementing such measures, and it is Kagi.

ak42

Vlad
about which blocklist to use, why not just allow users to select their own regex based blocklists in settings ? with compatibility with adblocker blocklist formats.
and have the browser check regularly for updates and store the lists in the local storage or something like that ?

Vlad

9rqnq I think this thread raises a question about somewhat different topic, a mechanism to report a new site as spam.

We could use one of the existing malware site lists, although I think known spam sites already on those lists appear rarely in Kagi to begin with.

dmvscam

I'm a new kagi user. Today I searched for "dmv opt out" and you can see the results below. Unthinkingly I clicked through the first result and submitted my information. A few minutes later, in horror, I realized this is a scam website. The FTC won a case against them in 2022, I don't know why the site is still up, but it certainly shouldn't be the top result on kagi. Now I expect to receive a flood of scam mailings and calls to my personal address and phone number 🙁 Please delist dmv.com, and all affiliated OnPoint scam websites, from your index.

FTC case: https://www.ftc.gov/news-events/news/press-releases/2022/04/ftc-win-at-trial-against-on-point-global-makes-102-million-in-refunds-for-consumers

OnPoint Global LLC list of scam websites: https://www.dmv.com/sites-list

Vlad

dmvscam What is the list where dmv.com is marked as spam?

Pat

Occasionally I stumble upon a malicious search result. An example is the third result of this search:

It's one of those sites that redirects you several times until you land on a, "You've been selected as a winner!" sorts of scams.

It would be nice if there was a built in way to help others by reporting malicious sites

Vlad

Pat Any idea how would we do this inside the crystal orb popup? (it is a bit cramped already) Also what is the expectation what we would with these sites to prevent potential abuse? Is the "Block" feature not enough?

Too many questions but things are never simple.

Pat

Vlad An idea I had - please forgive my poor design skills

Notice small report link beneath website ranking adjustment:

This would change the crystal ball (I'm not so sure if the large title bar makes sense here):

I suppose different reasons could be:

Phishing
Malware
Scam
Illegal content

The heuristic to actually hide results after reports is out of my wheelhouse, but I might suggest looking at uBlock's ban list in the meantime as it originally caught the malicious link in the OP.

Another idea is not hiding results but displaying a warning after some threshold.

handsel

@Vlad, maybe don't let users to "mark as malicious" (easy to abuse) but try to integrate lists like https://www.cert.pl/en/posts/2020/03/malicious_domains/ ?

Vlad

Even if we had this mechanism, malicious sites would require manual verification and then it can be subject to subjective opinion. Plus a malicious domain can expire, a small business may buy it for legitimate needs and it would stay flagged. It just seems to be messy.

handsel

Vlad, no, you could just use the list I gave you. It's official Polish malicious domain list. It's used by Orange, NextDNS etc.
Really, it's what it's for - to use 😉

Vlad

handsel We already use a similar list and any malicious sites are downranked and marked in results. Have you found sites that are in this list that are not marked on kagi?

matkam

The top result for “qc 2.0” (https://kagi.com/search?q=qc+2.0) redirects to a malware/phishing site, only sometimes. I’m not sure why it sometimes shows the desired content, and sometimes redirects to malware.

Kagi should detect malicious sites and not include them in search results.

Database

Can't reproduce on desktop. There is a chance that the site is not malware but is being hijacked by malicious advertising. The site seems to be generic SEO stuff so it still probably shouldn't be the top result.