Comment by like-the-rest | Hacker News Reader

Comment by like-the-rest | Hacker News Reader

Parent Full thread

  "We only do so when ordered..."

This is still a problem and to mince words over the difference between "voluntarily" versus "when ordered" is dodging the crux of the complaint: Protonmail is not unconditionally private to law enforcement, and so, it is not actually private at all.

It's not impossible to make Protonmail less private, provided there are extenuating circumstances, therefore we know there are controls in place to selectively reduce the supposed privacy that may exist under certain typical use cases.

The primary dishonesty here (in this blog post) is to imagine that the reader would think it would be probable for Protonmail to "volunteer" user data to law enforcement. Volunteer in the sense that they eagerly fetch all open cases, and actively push relevant matches to authorities (doing their work for them, not unlike one of Best Buy's Geek Squad informants locating hard drives for FBI compensation in cash awards of $500 per). Such a premise would be egregious and beyond the pale.

Meanwhile, what they really mean is that user data is not just available for the asking. Written warrants must justify and force the request.

Is that any better? No. Not by much. Not in a world of gag orders and warrant canaries. Those in search of services advertised as "truly private" expect more.

So what is expected? That user activity is rendered largely inaccessible even Protonmail employees, no matter how badly they'd like to crack the security encapsulating user information. Even under duress and threat of hostility.

Otherwise, what does that little "decrypting..." prompt even mean? Not much, evidently.

Now that we know this is not the reality (which they do purportedly profess) Protonmail is just another webmail service. There is no real novelty to their claims of privacy. It's just theater.

I don't see how it's a revelation that governments are able to use warrants to force companies to provide data. And we already knew that Protonmail could serve malicious code if forced to, that's been a common criticism of their mail service from day one.

To the extent that it is possible on the web today to do clientside encryption, Protonmail does clientside encryption. Yes, it's bound by platform limitations around code signing and browser crypto. But in a broad sense, all of the apps you have set to auto-update (including privacy gold-standards like Signal and Wire) are vulnerable to malicious update attacks if law enforcement is able to compel them. That has always been true.

So what new information have we actually learned at this point?

The only interesting part of this story to me was the idea that Protonmail might go beyond their legal requirements in assisting law enforcement. That part appears to be false.

No one should be using a 3rd-party service, whether advertised as "truly private" or not, under the expectation that the parent company is going to just ignore warrants and gag orders.

Protonmail has taken the perfectly reasonable position they will comply with the laws of the jurisdiction in which they are located: Switzerland. Other web mail services generally take the same position, though often in different jurisdictions. To the extent laws differ between Switzerland and the home countries of other services, they are different, and this would be true EVEN IF their technology, and policy for handling such requests, was identical to those of other web services.

From their privacy policy: https://protonmail.com/privacy-policy

We will only disclose the limited user data we possess if we are instructed to do so by a fully binding request coming from the competent Swiss authorities (legal obligation). While we may comply with electronically delivered notices (see exceptions below), the disclosed data can only be used in court after we have received an original copy of the court order by registered post or in person, and provide a formal response.

If a request is made for encrypted message content that ProtonMail does not possess the ability to decrypt, the fully encrypted message content may be turned over. If permitted by law, ProtonMail will always contact a user first before any data disclosure. Under Swiss law, it is obligatory to notify the target of a data request, although such notification may come from the authorities and not from the Company.

ProtonMail may from time to time, contest requests if there is a public interest in doing so. In such situations, the Company will not comply with the request until all legal or other remedies have been exhausted. Therefore, not all requests described in our Transparency Report will lead to data disclosure.

As a user who is primarily concerned with privacy in the sense of not having my emails read in order to advertise to me, this is perfectly reasonable. If I were planning a criminal enterprise or legitimately feared persecution by my government for my activities, I might take a harder look at email providers. Even then, as long as PM can't read my emails, I'd say that's far better than currently using Gmail.

Are you claiming ProtonMail hands over message content to law enforcement? That contradicts this quote from the post:

>Furthermore, ProtonMail’s end-to-end encryption means we cannot be forced by a court to provide unencrypted message contents.

OP is in the wrong about message contents. Message contents were never compromised to my knowledge. The issue was specifically metadata like login times and IP addresses. Also possibly mail server logs, such as that a user sent a message to a Gmail account, although I'm not 100% certain on that last point. Once again, when using the proper settings in ProtonMail, the message contents cannot be retrieved without a brute force attack thanks to end-to-end encryption.

I think the real meat of the question is the extent to which they cooperate with law enforcement. Do they respond to requests to extract keys from the client? Or used compromised keys for select users? If the government orders them to build such a mechanism to their client then encryption doesn't help. Even things like tracking the dates that messages are sent and received can make users vulnerable to correlation attacks.

I stand by the claim that the only truly secure mail is to do your own encryption and decryption client-side with open source tools. Anything done by a 3rd party service should be assumed to be compromised.

If the government could force them to extract keys from clients, to utilize compromise keys, or build a mechanism into a silent update, what makes you think and other company would have a different choice? Wouldn't such a government pose the same threat to any email provider?

> what makes you think and other company would have a different choice?

A company that operates in a country where the government is legally forbidden from doing such things would be more resilient to attack through the judicial or law enforcement system. Granted, there's always the suspicion that surveillance maybe be conducted extra-judicially, but it's still better. I don't know of any such countries. I think the Apple v. FBI case was going to set precedence in this matter, but that never went to trial.

Nothing prevents you from attempting your own client side security.

This is in no way related to the issue at hand.

>That user activity is rendered largely inaccessible even Protonmail employees, no matter how badly they'd like to crack the security encapsulating user information. Even under duress and threat of hostility.

Yes, hence the end to end encryption

Reply on news.ycombinator.com