Search for everyone
kagi.com751 points by elashri 2 days ago
751 points by elashri 2 days ago
If anyone from Kagi is reading this: Now that you have search without an account, is there any chance we can finally get a DNS-based enforced safesearch option like every other major search engine has? It's been in the feedback requests since May 2022[1].
Anyone on an enforced-safesearch network can now go to kagi.com, turn SafeSearch off, and search for porn or anything else, which is just going to get kagi blocked completely on all those networks. If you add a SafeSearch-enforced alternative IP address, then Kagi can still be used.
[1] https://kagifeedback.org/d/3285-safe-search-dns-locking-for-...
What legitimate reason is there for stuff like that? Workspaces should not be enforcing policies like that - trust your workers - and parental nonsense shouldn't be a goal of an ethical business.
It can be useful at home for parents with kids as well. A decent number of people use Pi-hole, NextDNS, or some other DNS blocker service.
Either way, ultimately the result of not offering the option will more likely be getting blocked entirely, than workers being freed from the shackles of SafeSearch.
What's even the point of DNS based safe search? Anyone can change their DNS and use DNS over https, how does DNS safe search work in that instance?
Also blocking the search is pretty weird, what prevents me of typing the URL directly?
> Anyone can change their DNS and use DNS over https
Not if the machine is locked down. Blocking all search engines is not a feasible approach, though, hence the Safe Search mechanism.
> Also blocking the search is pretty weird, what prevents me of typing the URL directly?
That URL can be blocked at the router level, but one can still see cached search engine images when safe search is off.
Enterprise and school devices often configure browsers to not allow modifying DoH settings and configure the OS to not accept unapproved programs like sideloaded browsers that may not obey that device policy.
It can be used to keep bad content away from kids. DNS over TLS can be blocked on the network router / firewall and DNS safe search enforced by forwarding lookups to something like Adguard
> What legitimate reason is there for stuff like that? Workspaces should not be enforcing policies like that - trust your workers
It's not just porn, it could be a way to prevent successful phishing through employees filling out a malicious form or whatever.
"Trust your workers" is bad infosec.
Yeah people are dumb if they can get an Amazon gift card or a job offer, see what happened to Binance.
But blocking alternative search engines and AI engines goes a bit far IMO especially if they dont use training. Especially if you're just afraid they will see a nipple on a search engine. That's ridiculous.
Kagi is already blocked in my workplace due to their AI features, so a 'sanitised', search-only version of the site would be very useful
> Kagi is already blocked in my workplace due to their AI features
Can you expand upon this? Why is there an AI block?
So that people don't feed proprietary data into a model that will train itself from it. Standard DLP practice.
Usually it's "You can only use CoPilot"
Ya Microsoft lobbies for that, at our place too.
The reason is that copilot resells other people's models (they're just a middle man) and there's an easy way to cut them out. So they blow up the dangers of other people's AI. Whereas if your data is in M365 they have it all anyway whether you use copilot or not. So they're trying to build a moat so they don't easily get cut out.
> s there any chance we can finally get a DNS-based enforced safesearch option like every other major search engine has? It's been in the feedback requests since May 2022[1].
Not working with Kagi, but while they have RU contracts this will not be possible, can guarantee.
You mean Yandex/Russia? Yandex itself has an enforced SafeSearch IP (213.180.193.56) so I would have thought Kagi could too.
It does appear to me @HNReply99 just wants to watch pornography? You can do it man, although some on this site may suggest SICP instead, or few radical ones even suggest programming in Rust.
Haven't tried Kagi yet. I know many people are increasingly frustrated with Google. As a duckduckgo user honestly i typically find what i want, except I am feeling more and more like i don't recognize / trusts many of the top sites in a result. might be a perfect time to try it.
A few months ago I stopped paying for kagi and went back to DDG. I started noticing all the top results were AI generated, so I just went back to Kagi yesterday and it is a huge breath of fresh air. So much better!
I did something similar, when the Kagi Translate changes were announced earlier in the year. I tried a local hosted AI assistant (via openrouter) along with DDG/Brave.
80% of the time the search results were perfectly fine, but that small amount of time would royally annoy me and my breaking point was this: https://i.redd.it/r4q6y3vl5vdh1.png
So I went back to Kagi and appreciate it a little more (they've also rolled back on Translate to some degree).
I pay for kagi, but hardly ever use search any more.
Come to think of it, one of my boxes is not logged into kagi, and is mostly ai nonsense.
I guess I’ll keep paying them. :-)
"Top results" was the complaint, this hides AI coming from Duck but not from Duck's results.
> I know many people are increasingly frustrated with Google
I’m not sure if this holds true nowadays. I’d say it git pretty bad, but now Gemini summary pretty much covers all of my use cases. And for any meaningful query, it’s straight to ChatGPT/Claude.
Gemini summary is wrong way too often to be relied on, especially for anything niche
Very rare for me to use Google nowadays for something niche. For those cases, I would assume it requires multiple sources to get a definitive answer, so it goes to ChatGPT.
Ok but there’s still a high error rate, so why would you trust it with general queries? Especially if it’s something you aren’t already knowledgeable about
That is fair. I always ask for sources whenever it is something I am unfamiliar with instead of just trusting the response, that way I can go and double check and it ensures the agent at least read something on the topic
I've found the value of this approach has diminished over time - now I need to go and read the source and try to identify if it was generated by an LLM itself.
Why do you think there’s a high error rate for general queries nowadays?
Because of the high error rate on queries where I know the topic well enough to tell when the answer is way off base
Yeah fair. I guess I have a very different experience, as I mentioned above, I basically never use it for any niche/hard queries that depend on multiple sources or research. For all my common queries, it's reliable.
> For all my common queries, it's reliable.
Is it though? How often are you spot checking sources?