Hacker News Reader

Comment by latexr | Hacker News Reader

> When asked by The Post whether any government had requested a backdoor, Google spokesman Ed Fernandez did not provide a direct answer but suggested none exist: "Google cannot access Android end-to-end encrypted backup data, even with a legal order," he stated.

No, that does not suggest none exists, it only says they don’t have access to it. They could have chosen or have been ordered to give the keys to the government agency but not keep one themselves. I’m not saying that’s likely, just that it’s important to not take these statements as saying more than they do. They wouldn’t hesitate to use “technically correct” as a defence and you have to take that into account.

View on HN

Before people immediately think the worst of Google or other corporate representatives, be aware that people working in these companies need to weight their words carefully. From The Verge's article on the issue:

The UK has reportedly served Apple a document called a technical capability notice. It’s a criminal offense to even reveal that the government has made a demand. Similarly, if Apple did cede to the UK’s demands then it apparently would not be allowed to warn users that its encrypted service is no longer fully secure.

How does this work wrt false advertising laws? If I relied upon their end to end encryption and it turns out to be false advertising because there's a secret backdoor, who do I sue?

no one, you'll be in secret prison before you somehow gain standing

If the back door was used the a three letter agency sure.

If the backdoor was exploited by a criminal though?

so in this hypothetical, some blackhat is able to download data in mass from apple servers? And you're worried that the only thing stopping that, or creating a duty to protect the data is encryption?

Which is exactly why I’m making this point. If no government had requested a backdoor, they could’ve simply answered “no”. When you have to weight your words, it means you’re not at liberty to say whatever you want. That is itself a signal, and why warrant canaries are a thing.

https://en.wikipedia.org/wiki/Warrant_canary

You're right to point out how carefully worded these statements are. But I suspect it's rare for companies of Google's status to not have been asked for a backdoor. It's not really an informative question to ask Google.

Can you elaborate on why you say it is not informative?

My guess is Google, Microsoft, Signal, Apple, Cloudfare, etc etc etc have all been asked if they could make backdoors. I expect they have all been asked. It's not the same as asking if they have made a backdoor.

So I think a journalist asking an organization like Google if they've been asked isn't really informative, because they almost certainly have been.

I'm not sure how it's relevant other than to say an answer from Google's response might seem oblique, but they're also being asked obliquely and that colors how you might interpret their response.

Presumably because the answer is "of course yes".

Of course they were asked. That doesn’t matter, my point is the author is assuming more from the reply than what was said.

It’s like if you conspired with your brother to steal from the cookie jar. He stole the cookies while you distracted your parents. Later on your mother reports to your father:

> When asked whether they stole from the cookie jar, derbOac did not provide a direct answer but suggested they didn’t didn’t know who did it: "I did not see anyone removing cookies from the jar," they stated.

Your statement is factually correct, but it doesn’t say what your mother concluded.

That concept has always sounded like tech people trying to hack the law without the proper real-world legal knowledge, IMO.

Bruce Schneier wrote in a blog post that "[p]ersonally, I have never believed [warrant canaries] would work. It relies on the fact that a prohibition against speaking doesn't prevent someone from not speaking. But courts generally aren't impressed by this sort of thing, and I can easily imagine a secret warrant that includes a prohibition against triggering the warrant canary.

Lots of similar discussion on HN already, e.g. in https://news.ycombinator.com/item?id=5871541.

Simply answering "no" when that's the truth could be illegal too. The ability to say no creates the ability to say yes as well. If I ask Apple whether they got an order and they say "no", then a year later they say "we cannot confirm nor deny", well then that's a yes.

Kinda depends on judicial interpretations of free speech, but that's how warrant canaries work. Are warrant canaries legal in the UK? They seem to be in the US but idk how well established that is.

But they can still notify the public, through those canary statements. (I forgot the name commonly used).

For example (a simplistic one), you can have a statement like "we do not have any backdoors in our software" added to your legal documents (TOS, etc). But once a backdoor is added, you are compelled by your lawyers to remove that statement. So you aren't disclosing that you have added a backdoor. You're just updating your legal documents to make accurate claims.

Such actions, even just the act of deleting text, conveys a message you were ordered to not convey and the government is not likely to take too kindly to that.

That is a fraudulent TOS if you're lying to the customer though

Not exactly the same but I've had a discussion around a similar topic with a Canadian immigration lawyer. We were put in contact by a mutual friend and the lawyer was looking for an email provider that was hosted in Canada and didn't rely on any US-based services (e.g. spam filtering). I asked him about the requirement and he pointed out that it was legally impossible for him to simultaneously comply with the USA PATRIOT Act and Canadian data protection/privacy laws. The US Gov't could compel his email provider to disclose solicitor-client privileged data with a gag order, and by not telling his client he would be breaking Canadian law.

By putting statements like what you're proposing in your TOS or marketing material you are potentially setting yourself up for a situation where it's now impossible to comply with all applicable laws. As others have mentioned, Australia passed legislation preventing you from disclosing the existence or non-existence of specific legal documents; they're at least warning you up front that the canary itself is illegal. The solution is to not make marketing statements that would become fraudulent in a situation where you can't legally retract them, unfortunately.

Edit: since lawyers are mentioned here... if the lawyer who is telling you that you need to remove the line from the TOS is the same lawyer who told you it was ok to put the line in the TOS... you should probably find a new lawyer because they didn't think through the consequences of approving it in the first place.

> if Apple did cede to the UK’s demands then it apparently would not be allowed to warn users that its encrypted service is no longer fully secure.

One would think this runs afoul of other laws though, truth in advertising and similar.

Its such a legal minefield, and the UKs request borders on violating the sovereignty of other nations I can't see Apple complying, but maybe that's hopium talking.

> No, that does not suggest none exists, it only says they don’t have access to it. They could have chosen or have been ordered to give the keys to the government agency but not keep one themselves.

The whole definition of "end-to-end encrypted" is that only the two ends have the keys. If anyone or anything other than the two ends (the one sending and the one receiving) has access to the keys, it's not end-to-end encrypted.

Whatsapp has had end-to-end encryption since 2016. But it only added encryption to cloud backups in 2021. They didn't share any key material with Google, just backed up the messages and media without any encryption to begin with.

Yes exactly. Google is very careful to say that "Google cannot access Android end-to-end encrypted backup data" and notice it doesn't say that all Android backups are end-to-end encrypted. For what we know, Google could have decided to use non-end-to-end backups in the UK and end-to-end backups everywhere else.

I think that's the implication, not the definition. Data remains encrypted even when a third party gets access a key.

But if they could give a key to the government agency, it wouldn't be end-to-end encrypted, right? Or are you thinking they would have a copy of users' keys that they gave out? (Which I guess is technically possible.)

They could also cripple user key-generation. E.g. they choose random primes from a known subset. It would make communication crackable while also being difficult to detect.

It would be no different from how multiple devices and users access the same content (chat, shared data, etc.). The government’s keys would always be included in set which encrypts the real key. They don’t need the users’ key, Apple doesn’t need their private keys. So technically still end to end encrypted, just with a hidden party involved. Users have no way of knowing this doesn’t already happen.

And when their key leaks, it’s as good as no encryption, but still end-to-end encrypted.

If the other end is the government, then it's kinda valid? =)

You can not use a DH key exchange, and create the symmetric key by some procedure that is predictable, or encode the symmetric key with the government's public key and send it to them.

It doesn't stop being end-to-end when you add another end. We often do group chats that way.

Or you can create a side-channel and send al the data there. That would stop it from being end-to-end.

if google were to transfer the keys elsewhere, they would have (temporary) custody of the keys, granting them access, and invalidating the statement.

My layman’s understanding is that a user’s private key is used to decrypt a random key, which is then used to protect data. Shared files then only require adding key access to that small secret by someone who knows the original key. If one of the original public keys is always one held by authorities, Google never needs to have custody of the private key and can’t access the data themselves making the statement true, but misleading.

> they would have (temporary) custody of the keys

No, they would have had custody of the keys. Meaning it would still be true they cannot (now) access the data.

Reply on news.ycombinator.com