Google Chrome Pushes For User Protection With 'Not secure' Label (axios.com)

In an effort to force websites to better protect their users, the Chrome web browser will label all sites not encrypted traffic as "Not secure" in the web address bar, Google announced Thursday. From a report: Encrypted traffic allows users to access data on a website without allowing potential eavesdroppers to see anything the users visit. HTTPS also prevents meddlers from changing information in transit. During normal web browsing, Google currently displays a "Not secure" warning in the next to a site's URL if it forgoes HTTPS encryption and a user enters data. Now the browser will label all sites without HTTPS encryption this way.

Comments

Entire internet doesn't need to be https

[iamhassi]

This is silly. Google is saying every website needs to be https. That's not true and is a waste of money and time to make every site https

Re:Entire internet doesn't need to be https

[XanC]

First, I think yes, it does. Otherwise it will be snooped or manipulated.

Second, they're just making it clearer when a site isn't https. Not saying every site needs to be secure.

Re:Entire internet doesn't need to be https

[ArchieBunker]

Don't the fed have all the SSL master keys anyhow?

Re:Entire internet doesn't need to be https

[nitehawk214]

So you don't mind a 3rd party knowing the content of each webpage you have visited?

Re:Entire internet doesn't need to be https

[Richard_at_work]

Every site has *something* to lose - if it's not user credentials or personally identifiable information, then it's reputation or simply the ability for a third party to inject ads or crypto mining scripts into the page.

We have all seen the fall out of ISPs injecting ads into pages - Comcast and others have done it - so if you want to be *certain* your page reaches your audience as you intend them to receive it, http is no longer good enough (and hasn't been for years).

Re:Entire internet doesn't need to be https

[Anonymous Coward]

nope... except for pages where I actually log in I couldn't care less. There should rather be a warning if there is any 3rd party content, like AdWords or Analytics...

Re:Entire internet doesn't need to be https

[Anonymous Coward]

Short answer; Yes.

Long answer; hell Yes... except all those self-signed certs chrome/google seem dead set on crippling even more for browser use.

Re: Entire internet doesn't need to be https

[Anonymous Coward]

With your browser trusting 600 CAs by default it certainly has absolutely no value without DNSSEC and DANE.

Re:Entire internet doesn't need to be https

[thegarbz]

This is silly. Google is saying every website needs to be https. That's not true and is a waste of money and time to make every site https

They are doing nothing of the sort, they are only finally putting HTTP in the correct light: It's not frigging secure and never has been. The fact that so far we have put more effort into poorly encrypted but none the less far more secure than HTTP.

It made no sense. This finally does.

Re:Entire internet doesn't need to be https

[Anonymous Coward]

HTTPS security doesn't matter if I don't trust the content anyway. (I could be looking at https://sloashdot.org/ for example. Or even the genuine slashdot.org and it could still be utter nonsense. It really only matters for the small handful of sites that I visit where the identity of that site would make a material difference to me (bank, tax dept).

Given that, manipulation is a non-issue. I could be looking at manipulated version of slashdot and I wouldn't trust it any more or less. Snooping is a bit of a concern; but I suspect they get that anyway. (Besides, knowing the IP is 90% of it.)

Second, they're just making it clearer when a site isn't https. Not saying every site needs to be secure.
They absolutely ARE implying that every site needs to be 'secure'. By having 'secure' (and I suspect it will have some big red text or something) they will imply that it is a bad thing. They are wrong, it's far more nuanced than that.

Finally, https doesn't guarantee security. https://www.enteryourcreditcardscam.biz/ is "secure" - all that https protects is you talking to the web server. From there, who knows, it could be uploading your CC data to dropbox for all the web browser knows. It's not good that Chrome gives users a false sense of security.

As for snooping, well, it's a bit rich of Google -- who the hell runs Google AdSense and analytics? All those javascript files 'secure' under https? They (Google) are already snooping on you - just with consent of the web site owners.

Maybe that's it... Google doesn't want ISPs getting their hands on their juicy advertising revenue? Or they think security is "user to site" without realising it's the site itself?

Re:Entire internet doesn't need to be https

[Anonymous Coward]

that's really an nginx flaw (it shouldn't have to restart to update certs).

Re:Entire internet doesn't need to be https

[XanC]

It doesn't. It can gracefully reload.

Re:Entire internet doesn't need to be https

[tlhIngan]

It isn't even an issue of money either. Let's Encrypt offers free certificates so I don't want to hear that it is a time and money issue.

It's a reputation issue. Given Let's Encrypt has issued over 14,000 paypal phishing certificates, one would think you should revoke Let's Encrypt certificates. After all, if Symantec, Comodo or others issued those, we'd be calling for blood.

The only reason we aren't is because Let's Encrypt has big names like EFF and Mozilla behind them. But all the scammers are basically dragging them through the mud - are your EFF donations being used to scam poor old ladies out of their money? Is scamming people really the goal of EFF and Mozilla?

Heck, it's actually kind of funny because a new exploit opened up on sites using Let's Encrypt, because they have a well-known directory that's being used to hide cryptocurrency miners and other things, too.

Maybe if there was a way to grade the quality of a certificate - Let's Encrypt can be made low, sites that charge with a real valid billing address (i.e., used a credit card, as opposed to bitcoin) can be higher rated because there is accountability down the line - including down to a real name and address.

Re:Entire internet doesn't need to be https

[amorsen]

I generate my own key and use letsencrypt to certify it. The key does not leave my server.

The feds can force any number of certificate authorities to generate a certificate that matches mine, with a new private key. They can do exactly the same if I had a self-signed certificate.

They cannot, without doing a targeted attack and breaking into my server, get the actual private key that my site uses. Again, precisely the same as a self-signed certificate.

There is no security advantage to using a self-signed certificate.

Re:Entire internet doesn't need to be https

[amorsen]

There is a way to grade. If you want actual validation, you need an extended validation certificate.

Any other type of certificate is just a way to scam you out of your money -- they do not verify anything except the fact that you aren't piss-poor. If you think a car charge provides any verification, I give you How to use prepaid debit cards.

If anything, it should be forbidden to charge money for a certificate that isn't extended validation. However, with Let's Encrypt available, the market hopefully sorts it out.

Re:Entire internet doesn't need to be https

[KixWooder]

It prevents your ISP from injecting crap into your pages, like Comcast has been known to do.

Re:Entire internet doesn't need to be https

[Actually, I do RTFA]

Sniffing is a minor concern. The bigger problem, by far, are third party tackers. This is more an attempt by Google to monopolize tracking data than preventing it.

Also, it only protects knowing which specific page I visit on a site (they can tell from the IP address what website I'm visiting, right?). And that's unnecessary on many or most sites. On , WebMD pages matter, but when you go to XKCD?

Re: Entire internet doesn't need to be https

[Anonymous Coward]

Extended validation is a sham too. I got an EV code signing certificate recently for signing windows drivers. The only verification was that the CA called my prepaid phone number to ask if I am indeed a hardware engineer working for xxx. I said yes, and got the certificate. I could easily have lied.

Re:Entire internet doesn't need to be https

[hawkinspeter]

systemctl reload nginx does the trick for me.

Re:Entire internet doesn't need to be https

[Cajun Hell]

One of the obvious problems with this whole thing, is that what https does is somewhat more technical than the kinds of things laymen know about, and Google wants to "dumb down" the distinction in the UI to something succinct. So they chose one single word, "secure" instead of "this conversation is believed (to a somewhat degree of confidence) to be though party X's webserver (or with them plus other parties that they consented to be included), and oh by the way, we also encrypted it too."

It's "wrong" but such is the cost of brevity. Some people will get the wrong idea what it means, but Google's betting that overall, on average, laymen users will be generally better informed by the change.

Re:Entire internet doesn't need to be https

[lactose99]

Nope, reload, and it can be done online

Re:Entire internet doesn't need to be https

[Anonymous Coward]

Second, they're just making it clearer when a site isn't https. Not saying every site needs to be secure.

Then what was the point of phrasing it as "Not Secure"? That you should feel good knowing the site is "insecure"? That you shouldn't think that Clueless Users won't suddenly say "why isn't it secure" and start demanding "security"?

They phrased it that way precisely to make users do that. Encrypt All Of The Things!!!! Never mind that doing so renders using an alternative DNS utterly impossible what with altering the domain at all would break the TLS verification, and ICANN deciding to basically allow any TLD to be registered. (No blank TLD could be used by the alt DNS servers.) Breaks interop with older equipment, (have fun getting that old appliance to use modern TLS), creates a massive need of cert issuing for services that never used nor needed it previously, and will futher segment and isolate the internet as a whole for very little gain, that most people won't even notice, let alone know how to take advantage of. (Hence the branding by Google, $10.00 says we get a story in the future about how much Google has "helped" internet security advance by doing this.)

Also: HTTPS != Secure

Yes, I know what the 'S' in HTTPS means, No, that definition is wrong. It should be 'HTTPE' (Encrypted) not HTTPS (Secure). There is no security in the underlying certs. They are all signed by a bunch of third parties that most people will never verify. The certs are all so convoluted that determining if it's changed, is a non-starter for the general public. The certs rely on being installed automagicly to the point most people don't even know they exist, and therefore can't even begin to make a decision on, let alone one about trust. The certs all depend on DNS, which can be intercepted and changed. (Never mind someone wanting to use an alternative DNS. So yay, more net segmentation and isolation.) The whole thing is broken from start to finish as far as trust is considered, and without trust, there can be no security.

Also, this is Google messing around again. The same Google that loves changing the meaning of "Trusted Certificate" every 5 seconds, rendering the idea impossible to keep up with, (and costly), for both site operators and users alike. The same Google that loves not trusting anyone, especially their users, and constantly making "improvements" that break the user's input on the decision of trust. The same Google that decided SSL/TLS interception of user data (banking, government, medical, personal, etc.) was perfectly fine, but intercepting the user's search queries, or the system's updates was a non-starter. The same Google that decided that Android apps should be able to ignore administrator-installed certs, and even not need to allow the user to even SEE what was considered trusted by them.

So no, Google is the last idiot I want being backed on this issue. They are in no position to be dictating what is secure and what isn't, they've abused that power to a ridiculous level and have proven that they cannot be trusted themselves.

Re:Entire internet doesn't need to be https

[AmiMoJo]

The certs that Let's Encrypt issues don't certify identity. If you are assuming that they verify the identity of the site owner you made a mistake.

Let's Encrypt check that the key belongs to a person with the ability to edit the site. That's it. You can be reasonably sure your communications with that she can only be read by people who can edit the site, that's it.

Re:Entire internet doesn't need to be https

[DCFusor]

No point modding up a +5. How about the 30 or so websites on my LAN of things? No internet involved at all. Faking certs for that would be just plain stupid. Seems the whole world is set up for stupid consumers only - no one creates content, right? That's why asymmetric DSL is such a hit, right?

Re: Entire internet doesn't need to be https

[Kjella]

With your browser trusting 600 CAs by default it certainly has absolutely no value without DNSSEC and DANE.

All it takes is a few webmasters to take note of their security certificate fingerprint and check it from a random home/mobile connection or proxy and you'd see alarm bells go off if someone was trying to MITM the world. With HTTP they can just snoop on a fiber optic cable and nobody would know. So when it comes to protecting everyday people visiting everyday sites I think it has an effect.

Re:Entire internet doesn't need to be https

[amorsen]

Someone did not grant me a certificate. They simply signed my public key, certifying that they believe that my public key belongs to me. Whether you choose to believe them or not is immaterial. No one is able to make my security WORSE by signing my public key -- that is pretty much the basis of public key cryptography.

(TLS is broken in that it only allows one entity to sign a given key in a certificate. It is incredible that no one has fixed that yet.)

Re:Entire internet doesn't need to be https

[TheRaven64]

You don't seem to understand how TLS certs work. The encryption and the signing are different parts of the security model. If I want to provide TLS connections, I generate public and private key pair. The private key is basically a random number that only I know. The certificate is a combination of the public key and some information (for example, my organisation name, the relevant domain name, and so on). When I ask a CA to sign the certificate, the create a cryptographic signature from their private key and my certificate.

The encryption happens using my key pair. The client receives my public key and uses this to encrypt traffic to my domain. The decision to trust the certificate rests with the client, not with me. The fact that a CA has signed the certificate says that the CA believes that it applies to my domain (and, with an EV cert, to my organisation), but the client doesn't have to trust that: it's just another piece of information that they have available.

For a corporate Intranet, you probably want to set up your own CA and install its public signing certificate on all corporate machines, so that you know that no one can forge the certificates.

Re:Entire internet doesn't need to be https

[TheRaven64]

The two concepts are separate in TSL. The encryption and the certificate verification are entirely separable concepts within the protocol and within most implementations.

In use, they are usually conflated because encryption by itself is meaningless. As a client, I care that I have a secure connection to a specific server. A secure connection to somewhere random, which may or may not be the server that I expected, is not a secure connection in any meaningful sense.

Re: Entire internet doesn't need to be https

[Maritz]

has absolutely no value without DNSSEC and DANE.

"If it isn't perfect, it's worthless" - This AC

Re:Entire internet doesn't need to be https

[TheRaven64]

Let's Encrypt implements the ACME protocol. You are free to choose any client for the ACME protocol that you wish. If you don't trust any of the third-party implementations then you are free to either use theirs or write your own, but unless you're using your own implementation of the HTTP, TLS, and TCP/IP protocols for the rest of your web server then you really can't complain that you need to use an implementation of another protocol from a third party.

I prefer to use acme-client than the certbot (the EFF / Let's Encrypt recommended client), because it's written by paranoid people and runs as a bunch of processes with privilege separation between them. It has fairly simple configuration, including a deploy script that is run after fetching the new certificates, so you can move them to the correct locations and restart the relevant services.

Re: Entire internet doesn't need to be https

[PhYrE2k2]

LetsEncrypt is not a low-grade certificate. It is a domain validated certificate. It offers the exact same encryption option as any other certificate does. If anything, a shorter renewal period is an improvement to turn over a compromised certificate faster. It does not offer low security, and labeling it as such is incorrect.

EV certificates are a way to sell trust, but they sadly do very little to actually verify the company. A fake document later and you have your certificate. Plus youâ(TM)re assuming you trust all of the 20+ companies issuing those certificates. Business model of any paid certificate is a Total waste of money for businesses and individuals.

And what about individuals and small businesses. Why should a small online store or a personal website with a contact form or a payment page of a small contractor be any less trusted than a banks web site who paid $1000usd/yr for someone to look at their website or article of incorporation?

Re:Entire internet doesn't need to be https

[houghi]

One I really visit often is http://dataserver/ with the IP address 192.168.1.27. So please tell me how I should turn that into https without Google Chrome yelling at me that it is insecure.

Sure, I could try to use a name like https://dataserver.example.com... and use the external address instead of the internal one, but that makes it LESS secure.

Not possible with https://certbot.eff.org/ and I am not going to pay for it, if it is possible at all. Yes, I know I cab create my own signature, but Choogle Chrome does not trust it.

Not going to help

[DarkRookie]

Normal users are not going to see nor are they going to care.

Re:Not going to help

[Richard_at_work]

Why have I been modded as a troll? Examples of both of these issues have been shown in the wild - Comcast has injected ads and other things into third party web pages before, and crypto miners have been included on pages via ads or third party scripts, so it's only a matter of time before they are injected directly.

So why the troll mod? Every site has something to lose - reputation and users. HTTPS prevents your ISP or VPN provider from doing this.

That's a bug in DNS policy, not CA policy

[tepples]

Someone who shouldn't be allowed to have a certificate for bankofarnerica.com shouldn't even be allowed to own the domain bankofarnerica.com in the first place. Typosquatting is in the bailiwick of the UDRP.

Only if a server has a FQDN

[tepples]

Let's Encrypt makes it easy and free for every website to be https.

This is true of public websites. It is not true of private websites hosted by web servers on a home local area network. Examples include the configuration interface of your router or printer. These have no certificate because they have no fully-qualified domain name (FQDN).

Or is everyone who operates a LAN at home expected to already own a domain?

Re:Only if a server has a FQDN

[kqs]

I'm confused: are you saying that it is a problem if your printer config page says "not secure" in the browser bar?

GP should have said "every website that Google will index" rather than "every website", but that seemed understood to me.

Re:Only if a server has a FQDN

[LesFerg]

It gets annoying whenever I access a local device on my network and chrome presents it's warning page, then I have to click on a link to expand some extra text, which has a link to let me continue to the intended destination.

They should at either have a maintainable list of sites I deem trusted, or be able to recognize local network devices and shut the fuck up when I am accessing them.

My Octoprint service is one example. It runs on a raspberry pi on my workbench and I use it's web interface from my PC or phone frequently. I would rather not have to fuck about with chrome warnings when I just want to see my printer status.

The sad thing is I am starting to prefer other browsers which don't have these annoying features.

Re:Only if a server has a FQDN

[tepples]

They should at either have a maintainable list of sites I deem trusted, or be able to recognize local network devices and shut the fuck up when I am accessing them.

The latter leads to security failure, as your browser would trust "local network devices" operated by an attacker on the open WLAN at a coffee shop.

The sad thing is I am starting to prefer other browsers which don't have these annoying features.

Which might these be? The same features you decry in Google Chrome are likely to show up in other derivatives of Chromium, and Firefox is implementing the same features.

Can phishing be stopped at the domain level?

[tepples]

GoDaddy, Gandi, Namecheap, and other registrars have registered over 14,000 paypal phishing certificates. Should we call for registrars' blood too?

Trusted computing

[Impy the Impiuos Imp]

"WARNING! Secure label is inaccurate and does not apply to google.com, facebook.com, youtube.com, or any other giant site with backdoors for government monitoring as part of the Prism panopticon."

"WARNING! Does not apply to any website run on computers with Windows, with backdoors for government."

"WARNING! Does not apply to any computer with hardware from the US or China, with special chips or standard chips with backdoors for government."

"Don't worry, they won't abuse it, even though human history has no examples where it isn't abused by those in power against their political opponents to remain in power."

Re:Trusted computing

[kqs]

When you say "they won't abuse it", are you talking about some specific "they"? Or just a general whining that there is no perfect security plus everyone in power sometimes acts like shit? Cause I think we all agree with the second one already.

Then who offers the free domains?

[tepples]

It isn't even an issue of money either. Let's Encrypt offers free certificates

Only to a domain owner. Neither Let's Encrypt nor any other CA included in the browsers' default certificate store offers any certificates for use with (say) .local, the TLD reserved for use with multicast DNS. What certificate should (say) the configuration interface of your home NAS use?

WWw

[Cajun Hell]

Perhaps the best way to explain this would be to say that Chrome is merely a World Wide Web browser, not a general web browser. ;-)

That's the problem, it's a lie. Totally false

[raymorris]

If it said "not encrypted" that would at least be *true*.

Marking sites as "not secure" vs "secure" based on using HTTPS is simply a lie. The usage of HTTPS is only slightly correlated with security. It's the equivalent of labeling people "tall" if they're black, and "short" if they are Hispanic. In general, the average height of Hispanic people tends to be lower than the average height of black people, but assuming someone is tall because they are black is stupid, and the label would be misleading almost as often as it would be accurate.

Many, many sites infected with all sorts of malware are served up via HTTPS, and many perfectly safe sites have are just fine with http.

Labeling one "not secure" is a falsehood, but worse is that it implies those without the "not secure" label must be "secure", which is a *dangerous* lie.

Re:That's the problem, it's a lie. Totally false

[RonVNX]

That's Google. Google has a huge problem with facts and truth, and dumbing everything down to the point of being counter-productive.

W3C Candidate Recommendation: Secure Contexts

[tepples]

are you saying that it is a problem if your printer config page says "not secure" in the browser bar?

I'm saying it's a problem if I can't, for example, view media that I have stored on my NAS box because its presentation in the browser relies on JS APIs that are reserved for secure contexts.

Overstated

[Ungrounded Lightning]

"Don't worry, they won't abuse it, even though human history has no examples where it isn't abused by those in power against their political opponents to remain in power."

While this may be true, it is something of an overstatement - because you can't show it to be true for recent stuff. It take a while for info to leak out.

Make it something like:

Don't worry, they won't abuse it, even though human history has no examples (more than 30 years old) where it wasn't shown, within 30 years after the event, that it had been abused by those in power against their political opponents to remain in power.

and it might work.

Re:Grocery shopping with armoured guards ?

[thegarbz]

Are you shopping for your groceries using an armoured truck with a set of guards carrying heavy weapons ? If not, why ?

I'll tell you what I'm doing. I am shopping for my groceries with a truck and security detail that is expressly made clear to me.

When I go grocery shopping in my armoured truck I know it's secure like a Secure HTTPS certificate.
If I go grocershopping and someone in my security detail is on the take the security company will make that known to me and inform me when I'm insecure, just like a breached HTTPS trust.
When I go grocery shopping in my minivan without security I know about that too completely fucking unlike how current browsers handle HTTP.