I had my personal domain I use for self-hosting flagged. I've had the domain for 25 years and it's never had a hint of spam, phishing, or even unintentional issues like compromised sites / services.
It's impossible to know what Google's black box is doing, but, in my case, I suspect my flagging was the result of failing to use a large email provider. I use MXRoute for locally hosted services and network devices because they do a better job of giving me simple, hard limits for sending accounts. That way if anything I have ever gets compromised, the damage in terms of spam will be limited to (ex) 10 messages every 24h.
I invited my sister to a shared Immich album a couple days ago, so I'm guessing that GMail scanned the email notifying her, used the contents + some kind of not-google-or-microsoft sender penalty, and flagged the message as potential spam or phishing. From there, I'd assume the linked domain gets pushed into another system that eventually decides they should blacklist the whole domain.
The thing that really pisses me off is that I just received an email in reply to my request for review and the whole thing is a gas-lighting extravaganza. Google systems indicate your domain no longer contains harmful links or downloads. Keep yourself safe in the future by blah blah blah blah.
Umm. No! It's actually Google's crappy, non-deterministic, careless detection that's flagging my legitimate resources as malicious. Then I have to spend my time running it down and double checking everything before submitting a request to have the false positive mistake on Google's end fixed.
Convince me that Google won't abuse this to make self hosting unbearable.
This may not be a huge issue depending on mitigating controls but are they saying that anyone can submit a PR (containing anything) to Immich, tag the pr with `preview` and have the contents of that PR hosted on https://pr-<num>.preview.internal.immich.cloud?
Doesn't that effectively let anyone host anything there?
Dark-grey text on black is cursed. (Their light theme is readable.)
Also, you can do bulk inserts in postgres using arrays. Take a look at unnest. Standard bulk inserts are cursed in every database, I'm with the devs here that it's not worth fixing them in postgres just for compatibility.
If you're going to host user content on subdomains, then you should probably have your site on the Public Suffix List https://publicsuffix.org/list/ .
That should eventually make its way into various services so they know that a tainted subdomain doesn't taint the entire site....
In another comment in this thread, it was confirmed that these PR host names are only generated from branches internal to Immich or labels applied by maintainers, and that this does not automatically happen for arbitrary PRs submitted by external parties. So this isn’t the use case for the public suffix list - it is in no way public or externally user-generated.
What would you recommend for this actual use case? Even splitting it off to a separate domain name as they’re planning merely reduces the blast radius of Google’s false positive, but does not eliminate it.
A friend / client of mine used some kind of WordPress type of hosting service with a simple redirect. The host got on the bad sites list.
This also polluted their own domain, even when the redirect was removed, and had the odd side effect that Google would no longer accept email from them. We requested a review and passed it, but the email blacklist appears to be permanent. (I already checked and there are no spam problems with the domain.)
We registered a new domain. Google’s behaviour here incidentally just incentivises bulk registering throwaway domains, which doesn’t make anything any better.
I think the other very interesting thing in the reddit thread[0] for this is that if you do well-known-domain.yourdomain.tld then you're likely to get whacked by this too. It makes sense I guess. Lots of people are probably clicking gmail.shady.info and getting phished.
Not sure if this is exactly the scenario from the discussed article but it's interesting to understand it nonetheless.
TL;DR the browser regularly downloads a dump of color profile fingerprints of known bad websites. Then when you load whatever website, it calculates the color profile fingerprint of it as well, and looks for matches.
(This could be outdated and there are probably many other signals.)
Tangential to the flagging issue, but is there any documentation on how Immich is doing the PR site generation feature? That seems pretty cool, and I'd be curious to learn more.
The one thing I never understood about these warnings is how they don't run afoul of libel laws. They are directly calling you a scammer and "attacker". The same for Microsoft with their unknown executables.
They used to be more generic saying "We don't know if its safe" but now they are quite assertive at stating you are indeed an attacker.
> The one thing I never understood about these warnings is how they don't run afoul of libel laws. They are directly calling you a scammer and "attacker"
Being wrong doesn't count as libel.
If a company has a detection tool, makes reasonable efforts to make sure it is accurate, and isn't being malicious, you'll have a hard time making a libel case
There is a truth defence to libel in the USA but there is no good faith defence. Think about it like a traffic accident, you may not have intended to drive into the other car but you still caused damage. Just because you meant well doesn't absolve you from paying for the damages.
I'm not sure what exactly you mean by "good faith". The legal standard is "actual malice" which means you had to have either known you were wrong or recklessly disregarded being wrong. Just saying something false about a public figure in a way that causes damages is not enough.
The first step in filing a libel lawsuit is demanding a retraction from the publisher. I would imagine Google's lawyers respond pretty quickly to those, which is why SafeBrowsing hasn't been similarly challenged.
I'm fighting this right now on my own domain. Google marked my family Immich instance as dangerous, essentially blocking access from Chrome to all services hosted on the same domain.
I know that I can bypass the warning, but the photo album I sent to my mother-in-law is now effectively inaccessible.
Maybe a dumb question but what constitutes user-hosted-content?
Is a notion page, github repo, or google doc that has user submitted content that can be publicly shared also user-hosted?
IMO Google should not be able to use definitive language "Dangerous website" if its automated process is not definitive/accurate. A false flag can erode customer trust.
> The most alarming thing was realizing that a single flagged subdomain would apparently invalidate the entire domain.
Correct. It works this way because in general the domain has the rights over routing all the subdomains. Which means if you were a spammer, and doing something untoward on a subdomain only invalidated the subdomain, it would be the easiest game in the world to play.
This happened to one of our documentation sites. My co-workers all saw it before I did, because Brave (my daily driver) wasn't showing it. I'm not sure if Brave is more relaxed in determining when a site is "dangerous" but I was glad not to be seeing it, because it was a false positive.
This is #1 on HN for a while now and I suspect it's because many of us are nervous about it happening to us (or have already had our own homelab domains flagged!).
So is there someone from Google around who can send this along to the right team to ensure whatever heuristic has gone wrong here is fixed for good?
A good takeaway is to separate different domains for different purposes.
I had prior been tossing up the pros/cons of this (such as teaching the user to accept millions of arbitrary TLDs as official), but I think this article (and other considerations) have solidified it for me.
98 comments
[ 0.17 ms ] story [ 124 ms ] threadhttps://old.reddit.com/r/immich/comments/1oby8fq/immich_is_a...
I had my personal domain I use for self-hosting flagged. I've had the domain for 25 years and it's never had a hint of spam, phishing, or even unintentional issues like compromised sites / services.
It's impossible to know what Google's black box is doing, but, in my case, I suspect my flagging was the result of failing to use a large email provider. I use MXRoute for locally hosted services and network devices because they do a better job of giving me simple, hard limits for sending accounts. That way if anything I have ever gets compromised, the damage in terms of spam will be limited to (ex) 10 messages every 24h.
I invited my sister to a shared Immich album a couple days ago, so I'm guessing that GMail scanned the email notifying her, used the contents + some kind of not-google-or-microsoft sender penalty, and flagged the message as potential spam or phishing. From there, I'd assume the linked domain gets pushed into another system that eventually decides they should blacklist the whole domain.
The thing that really pisses me off is that I just received an email in reply to my request for review and the whole thing is a gas-lighting extravaganza. Google systems indicate your domain no longer contains harmful links or downloads. Keep yourself safe in the future by blah blah blah blah.
Umm. No! It's actually Google's crappy, non-deterministic, careless detection that's flagging my legitimate resources as malicious. Then I have to spend my time running it down and double checking everything before submitting a request to have the false positive mistake on Google's end fixed.
Convince me that Google won't abuse this to make self hosting unbearable.
https://news.ycombinator.com/item?id=45538760
Doesn't that effectively let anyone host anything there?
Dark-grey text on black is cursed. (Their light theme is readable.)
Also, you can do bulk inserts in postgres using arrays. Take a look at unnest. Standard bulk inserts are cursed in every database, I'm with the devs here that it's not worth fixing them in postgres just for compatibility.
What would you recommend for this actual use case? Even splitting it off to a separate domain name as they’re planning merely reduces the blast radius of Google’s false positive, but does not eliminate it.
This also polluted their own domain, even when the redirect was removed, and had the odd side effect that Google would no longer accept email from them. We requested a review and passed it, but the email blacklist appears to be permanent. (I already checked and there are no spam problems with the domain.)
We registered a new domain. Google’s behaviour here incidentally just incentivises bulk registering throwaway domains, which doesn’t make anything any better.
0: https://old.reddit.com/r/immich/comments/1oby8fq/immich_is_a...
https://blog.chromium.org/2021/07/m92-faster-and-more-effici...
Not sure if this is exactly the scenario from the discussed article but it's interesting to understand it nonetheless.
TL;DR the browser regularly downloads a dump of color profile fingerprints of known bad websites. Then when you load whatever website, it calculates the color profile fingerprint of it as well, and looks for matches.
(This could be outdated and there are probably many other signals.)
They used to be more generic saying "We don't know if its safe" but now they are quite assertive at stating you are indeed an attacker.
Being wrong doesn't count as libel.
If a company has a detection tool, makes reasonable efforts to make sure it is accurate, and isn't being malicious, you'll have a hard time making a libel case
For instance: https://reason.com/volokh/2020/07/27/injunction-in-libel-cas... (That was a default judgment, though, which means Spamhaus didn't show up, probably due to jurisdictional questions.)
The first step in filing a libel lawsuit is demanding a retraction from the publisher. I would imagine Google's lawyers respond pretty quickly to those, which is why SafeBrowsing hasn't been similarly challenged.
I know that I can bypass the warning, but the photo album I sent to my mother-in-law is now effectively inaccessible.
https://news.ycombinator.com/item?id=42779544#42783321
Unironically, including a threat of legal action in my appeal on the Google Search Console was what stopped our instance getting flagged in the end.
Is a notion page, github repo, or google doc that has user submitted content that can be publicly shared also user-hosted?
IMO Google should not be able to use definitive language "Dangerous website" if its automated process is not definitive/accurate. A false flag can erode customer trust.
Correct. It works this way because in general the domain has the rights over routing all the subdomains. Which means if you were a spammer, and doing something untoward on a subdomain only invalidated the subdomain, it would be the easiest game in the world to play.
malware1.malicious.com
malware2.malicious.com
... Etc.
So is there someone from Google around who can send this along to the right team to ensure whatever heuristic has gone wrong here is fixed for good?
I had prior been tossing up the pros/cons of this (such as teaching the user to accept millions of arbitrary TLDs as official), but I think this article (and other considerations) have solidified it for me.
For example
www.contoso.com (public)
www.contoso.blog (public with user comments)
contoso.net (internal)
staging.contoso.dev (dev/zero trust endpoints)
raging-lemur-a012afb4.contoso.build (snapshots)