Hacker News Reader

Comment by LinuxBender | Hacker News Reader

This will be a taboo and unpopular option. My theory is that these discussions are theater. More often than not people store and access their data from their cell phones. Between CarrierIQ and OTA updates/access, there is no such things as end-to-end encryption on a cell phone. People get really upset when I bring this up. I suspect it is a matter of denial and not providing links to public documents, which will never exist. I would suggest that very few people have the patience to implement proper OpSec with their own data.

You don't even need "backdoors" in encryption. Existing lawful-intercept on Slack, Discord, Facebook, Google and all the wireless carriers will net just about anything you could ever want to know.

View on HN

> People get really upset when I bring this up. I suspect it is a matter of denial and not providing links to public documents, which will never exist.

The last time I saw this matter being discussed here, tptacek was getting pretty upset at people who were suggesting that baseband processors presented a security threat. He's a security expert with a reputation to uphold so I expect he's probably right, but I would still like to see a detailed explanation of why he's right.

For a reason I don't understand, this topic seems to illicit a lot of insults, when calmly assuaging fears would doubtlessly be more effective.

For those interested: here is one of those posts: https://news.ycombinator.com/item?id=14240626

Is there somewhere I could read some more on this point?

I was under the impression that iMessage was end-end encrypted. So unless Apple has a secret backdoor built into their systems, nobody should be able to access those messages correct? Wireless carriers are just sending my encrypted messages over their networks.

What am I missing?

Do you believe my cell service provider could decrypt my HTTPS connections to my bank?

I think the point is that the govt can sidestep https by just going to the bank; as they can to most services.

It doesn't matter if the apps try to be secure. If any point in the communication holds the key-data in memory and is not on a //user// trusted device then the key is compromised.

Since cellphones have a BMC that isn't user owned, which does have access to the full system memory (rather than being an isolated peripheral modem), and since OTA firmware updates can be pushed by the "infrastructure" (including fake 'towers' setup by TLAs, criminal hackers of other sorts, and hobbyists) containing code to compromise and silently ex-filtrate any data (including those keys, or even just the conversation directly); it is an inherently insecure environment.

What should the average security-conscious person do then? Resort to one-time pads? Is there any way to be truly secure, or should we just stop trying?

If we want secure tools for democracy then that's going to mean a completely open and thus verifiable (audit-able) platform.

This needs to be from the PCB traces, all of the component tolerances, all of the chips, all of the firmware (even the ROMs that are actually baked in ROMs on the chips), the bootloader, OS, and entire userland.

This is required not just for the host system but also the human interfaces and peripherals.

I hope we will be able to reach that point with a RISKV system at some point; but the various proprietary interfaces that require licences for implementation/etc might make this problematic. I am for standards, preferably completely free, but FRAND and non-restrictive on meeting the above goals might be good enough. The platform has to be fully open-book, but some of that book can be covered by reproduction limitations for a limited time. (I'd prefer standard patent duration at most, as this stuff NEEDS to become the digital version of paper at some point; and within my lifetime would be nice.)

Apple showed that they are capable and willing to perform silent over the air updates to operating systems to address security concerns. The specific incident I am referring to is with Zoom.

If you can read your message history from an iPhone, Apple is just one silent update away from reading it as well.

Not sure why this comment by robdachshund is dead: "Why bother trying to break encryption when you have backdoors in the hardware and OS? It's far easier to just use a keylogger on your OS keyboard, take a screenshot, access RAM buffers, etc."

To answer to that. Maybe the problem is in those hardwares and operating systems that do not have backdoors?

You're missing that Apple has the ability to put backdoors in kernel updates, and these can reach right into process memory and get your unencrypted messages.

I'm pretty sure he's just referring to the fact that if someone can push an ota update and you have something like iMessage that has encryption keys on the device, there's nothing stopping a company from pushing an update out that compromises your security. End to end isn't compromised, it's potentially compromised. Smartphones are the schrodinger's cat of security because while they're considered "secure" right now, they could be blown wide open via automatic update any second. Telegram is encrypted end-to-end, but if you have automatic updates on, there's nothing from stopping the developer or rogue actors (or states, like russia) from pushing an update out that uploads all of the encryption keys to their servers and decrypts everyone's messages. Automatic updates are fundamentally insecure because they can instantly compromise your security. If you want true security, go with open source and review (or wait for reviews) of all code changes that happen in the apps you use, and you want a distro that either doesn't have an os ota system or allows you to disable it. Also, there is potential for baseband attacks because we're all at the mercy of telecoms and no one runs custom firmware or opensource code on their baseband processor so who knows what vulnerabilities lie there.

> Between CarrierIQ and OTA updates/access, there is no such things as end-to-end encryption on a cell phone.

I don't think you understand what end-to-end means.

Even with end to end at some point the information needs to be displayed to the user and especially on Android at that point you can collect anything you'd like. Unless there is some way of encrypting data all the way to the display controller I don't know about. But that will never happen as it would be the end of so many other features.

So he's right theres no such thing as true end to end on common cell phones.

> no such thing as true end to end on common cell phones.

Hardly the only application though, is it?

Unless there is some way of encrypting data all the way to the display controller I don't know about.

That's kind of what HDCP is, so it could be done on an embedded display too.

IIUC you'd have to encrypt the text rendering, not the rendered text, for the message to be e2e encrypted and inaccessible through memory.

In theory you could run the decryption and rendering in the GPU (as is done for video), which would re-encrypt for display.

I most certainly do. CarrierIQ (former name) negates it. HTTPS and GPG can't hide anything from it. Any phone app will be entirely transparent. There are other debug apps embeded in different phones that can be triggered to start gathering data in the background. They can even tell the velocity you swiped in what direction and what angle you were holding the phone.

> I most certainly do.

I do not find that this comment proven, neither in your profile, nor in the context of this thread.

The article doesn't seem to indicate which devices are communicating, just generally referring to communication. Medical devices which communicate should certainly not be an acceptable communication to risk.

Reply on news.ycombinator.com