Hacker News Reader

Comment by p-e-w | Hacker News Reader

Unless I'm blind, the article doesn't contain any actual evidence of telemetry being sent – just a list of domains contacted, and speculation on what their purpose might be.

I don't doubt that Windows collects and transmits telemetry data (hell, text editors do that nowadays), but if an analysis of that exists, it is not in this article.

View on HN

> I don't doubt that Windows collects and transmits telemetry data (hell, text editors do that nowadays), but if an analysis of that exists, it is not in this article.

Such an analysis does not exist because that traffic is encrypted. Which is also the reason why using Windows 10/11 is not fully compliant with EU privacy laws in places like Germany, as there is no telling what Windows is actually phoning home.

Officially Windows 10/11 can be used but only after jumping through a lot of hoops that involve turning off the telemtry and phoning home, but even then only with an "acceptable residual risk" [0]

The only reason this isn't a bigger topic is because there is no realistic alternative; Everything is tailored to MS, and MS spends absurd amounts of money and effort to prevent anything from changing that.

So the majority just goes with the "easiest" and most convenient solution, even when it might actullay be an "illegal" solution that enables a ton of industrial espionage.

[0] https://www.heise.de/news/Datenschutzkonferenz-Hohe-Huerden-...

> Such an analysis does not exist because that traffic is encrypted.

... by software that resides on the same system, with keys that are in memory on the same system.

I'm not saying it's trivial to decrypt the traffic, but it's certainly possible, and much, much harder reverse engineering is routinely being performed.

Then go right ahead and do that, there will be a myriad of official government instutions, from all over the world, that would be very interested in your findings.

Or MS could simply share the keys with those government institutions there have been literally asking for it, to see wether Windows is actually sending home privacy relevant data.

But the matter of fact is it's a very real issue and still on-going problem.

Just because investing a lot of effort could shed some further light on it does not really change anything about that or the non-compliant behavior MS engages in.

Security only being as good as the effort lobbed at it to break it, is not really a novel or useful insight in this scenario.

What addresses are connected: https://learn.microsoft.com/en-us/windows/privacy/manage-win...

What data is sent: https://learn.microsoft.com/en-us/windows/privacy/required-d...

The Diagnostic Data Viewer is a Windows app that lets you review the Windows diagnostic data your device is sending to Microsoft, grouping the info into simple categories based on how it's used by Microsoft. https://learn.microsoft.com/en-us/windows/privacy/diagnostic...

Yes we know all that. What I want is a demonstration of live decryption of telemetry data traffic.

> with keys that are in memory on the same system

I'm not sure that actually holds — the encryption keys are in memory, but the decryption keys don't necessarily have to be.

The pre-encrypted payloads definitely are in memory at some point; however snatching them probably involves larger-scale reverse-engineering.

If they're using standard TLS, the actual data encryption is symmetric, so the encryption keys are the decryption keys and must be in memory during the encryption process.

If it is TLS you can get the keys used in the session from lsass’ memory. I’ve even written a tool to do so in PowerShell https://gist.github.com/jborean93/6c1f1b3130f2675f1618da5663.... This will generate a log file that contains the keys needed for Wireshark to decrypt TLS traffic.

My claim is it's not standard TLS or there's an additional layer (external encryption key) because an actual decryption of telemetry traffic has never been demonstrated.

>just a list of domains contacted, and speculation on what their purpose might be

Yeah, they might be playing some patty cake protocol /s

Yeah, this comes up over and over again: people see that connections are being made, but because it's hard (not impossible! but considerable work) to trace where they're coming from and what they actually _contain_, we're left with ambiguity.

(you'd need the Windows kernel debugger and to reboot the thing in developer mode. This is a pretty niche skillset)

Google - truly an niche skill.

What addresses are connected: https://learn.microsoft.com/en-us/windows/privacy/manage-win...

What data is sent: https://learn.microsoft.com/en-us/windows/privacy/required-d...

You posted this twice but i'll reply to the first one.

If you trust Microsoft to not send data you don't want to be sent why worry about telemetry and other potential spyware from them in the first place? You already have trust in them anyway.

But if you do not trust Microsoft to do that, then why trust their claims? They could be collecting a bunch more and if ever found they can claim that it was a bug like a bunch of other companies already did (e.g. HP having keyloggers on their laptops).

Impressive! Now the $64,000 GDPR question: is that all of them? And which of the long list might be considered personally identifying, including "jigsaw identification"?

(this is also vindicating the "the best way to get correct information on the internet is to post incorrect information"; all the replies could have been top level replies to the OP but weren't)

Windows 10 came out 8 years ago.

> This is a pretty niche skillset

Is it really? Seems you need to install ~2 packages, flip some switches, edit some files/run some terminal commands, reboot and connect the debugger and you're pretty much setup for it. Seems more like something you need to read an article about, rather than a skillset per se.

"You don't pay the plumber for banging on the pipe. You pay him for knowing where to bang."

Cute, I read the same submission yesterday too :) But, not really applicable in this case.

Then why isn't the output in any of these articles?

The author is a self-described "tech journalist" and it's not really news that most of them tend to be lazy and not really dive even surface-level deep into any topics. Author probably wrote the article with a deadline approaching fast, and editor said it was good enough, so published it became.

This video it's based on was from earlier in the year and lacks any proper analysis, yet got picked up by some tech sites and continues to get posted for topics which mostly just results in users bikeshedding about their Windows pet peeves.

Some examples: calls a service that helps companies comply with GDPR "1984" since the analysis is superficial and almost entirely gut reactions to random domains they observed. They have no issue with one CDN they observe but take issue with a Microsoft CDN literally only because they see 'geo' in the sub-domain, when if they'd taken a moment to check it's just used to locate the nearest data center like CDNs do generally.

If the video were instead, 'let's look at what comes pre-installed with Windows 11 [Home, Pro, Education?] that connects online' then it may have been a more informative video. Instead it has a lot of assumptions but little evidence about DNS queries, leaving only casual viewers thinking they've learned something.

You can easily transmit information hidden in web requests. If I own a web server at somedomain.com I can run code that asks for somedomain.com/someencryptedinformation.html, and I'm sure that someencryptedinformation will appear in the server logs. Automate encoding something in that string and you can send data using simple web requests that won't trigger anything looking for connections on non standard ports. And what if connecting to domain A followed by B within 2 seconds is translated as a 0 and the other way around as a 1? You can, albeit slowly, send whatever you want just by alternating those two connections. Yes, just contacting the outside is sending information.

>Unless I'm blind, the article doesn't contain any actual evidence of telemetry being sent – just a list of domains contacted, and speculation on what their purpose might be.

>I don't doubt that Windows collects and transmits telemetry data (hell, text editors do that nowadays), but if an analysis of that exists, it is not in this article.

Crikey mate; second paragraph, first sentence.

Yes and then if you read paragraphs 3, 4, 5, etc. there's no substance on what is actually sent. We all know what Wireshark is, and what it can do, but how it was utilized here is up in the air.

> there's no substance on what is actually sent.

Does Microsoft provide any transparency on what is sent?

What enhancement do you expect from the tecnical writer considering lack of sources?

I was replying to someone.

Reply on news.ycombinator.com