I think that’s uncharitable. Everyone is going through the motions required of them, and this is the public demonstration of those mechanizations (although Signal is a bit cheeky, which is fun). The next step would be government requiring, through legislation, more invasive logging and data collection (Australia and parts of Europe have already seen the beginnings of this discussion) of messaging apps (“we’ve asked for what we can, they said they don’t have it and aren’t required to have it, what do you want us to do?”).
When encryption and secure messaging is outlawed, only outlaws will have and use it.
With enough effort, anyone can go to jail. America held a taxi driver for 17 years at Guantanamo Bay with no evidence. Tech won’t save you from the state. As always, if your threat model includes a state actor, you are going to have a bad time. For all intents and purposes, their resources are unlimited.
Freedom is won in the courts and the legislature, not in the code (although tech is as useful tool for keeping government implantations in check).
(I still use and donate to Signal, but have a healthy understanding of its limits)
They can put one or two people in jail, but they can't put everyone in jail. If everyone has easy access to end-to-end encrypted messaging and relies on it (for non-nefarious purposes), the government will have a tough time changing that.
Yes, but their tyranny must also increase in order to circumvent the technology. They will increasingly resort to actions like you described. Hopefully the population will eventually revolt and put an end to the corrupt government once it becomes unacceptably totalitarian.
Freedom is won through weapons. Encryption is a potent weapon, it can defeat states, militaries. Before computers, it used to be a military tool. It must be democratized, the whole world must use it.
Isn't this what happened to Protonmail? They were required by legal order to start logging activity for a specific group of users. It's not outside the realm of possibility that the govt could try to force a company to either start logging Signal metadata or provide a backdoored app to a user. Not that it would necessarily work, but I do expect them to try at some point.
> When encryption and secure messaging is outlawed, only outlaws will have and use it.
They don't necessarily need to outlaw it. They may just throw up enough hurdles that it doesn't become a major success. Developing a communication system that is secure, featureful and convenient to use for the general population is not a trivial task. A large effort that can be undermined.
E.g. if they only require logging from communication service providers but not from application developers then this would force a decentralized solution. If they lean on payment providers it might get difficult to charge for phone apps or get donations.
The software could continue to legally exist but see little adoption. Which is enough to enable surveillance.
This isn't their first rodeo. The DOJ is well aware of what happens when they send subpoenas to Signal. They're not sending it because they're unaware of the probable result.
Am I the only one sort of bothered by the fact they shared that specific information with the world? It may not seem like much, but that was user data.
It's potentially useful to the target of the subpoena, who knows when they created their Signal account, and might thus now know that they're under investigation.
Not only all the records that they have, but it proves that the data isn't meaningful to de-anonymize someone. If they had to redact it we would wonder why and how that information would be useful.
The snark is publishing it in the blog post not blacked out. As a side effect the account may or not be warned by this. Not sure, if it’s legal to do so, in the US.
Speaking of donations (a guy from a food bank whom I see in the Safeway parking lot didn't know this, so I think we can assume not everyone does):
Most "donate" pages do not allow for "donor-advised funds (DAF)." They assume you're giving it with your before-tax money and presumably taking a tax deduction for it.
In a DAF, which your financial institution surely offers, you can donate appreciated assets, e.g. your FAANG stock, and take the entire amount as a tax deduction. So if your 10 shares of Facebook (excuse me, "Meta") stock are at 322, you can take a deduction of $32,200 this year.
What's the catch? That money's gone, and you can't get it back. You can only "advise" your DAF to give it to a 501(c)(3) organization, which Signal is. There are no time limits.
The good part, though, is you can probably have your DAF give the money anonymously, so the charity can't bug you every time they're having a fund drive.
Another benefit, it sounds like, is that you don't have to pay capital gains on selling those shares.
Like, let's say your intent is to donate $10k to some charity, out of the goodness of your heart and/or as a tax write off. You don't have that in cash, but do in stock.
You could liquidate $10k of stock, pay capital gains on it (if it appreciated since acquisition), then donate it. So you're out the capital gains tax.
The method you describe seems more efficient, since you don't need to sell; you simply transfer ownership of the asset.
Or is there still capital gains to be paid?
I wonder if billionaires are setting up charities as trusts for their kids, then "donating their shares to charity?"
You're exactly right, you don't pay capital gains tax, and DAFs really are the poor man's "tax-exempt foundation."
Billionaires have access to much fancier schemes than this, and I won't even attempt to describe all those. But yeah, I imagine "donating their shares without capital gains taxes" figures into them.
I just noticed you said "trusts for their kids" -- that's something different. If the children can access it, it's not a DAF. But trusts are much more complicated, and someone who understands them (which I don't) can hold forth here.
Yeah, I didn't realize what an enormous difference this made until I ran the numbers.
In your example above, let's say the person purchased those 10 Meta shares for $38 each at the IPO and they're worth $322 each now. That's $3220 in proceeds and a $2840 capital gain.
The taxes on this depend on income level and state of residence, but let's say they're in CA making $300K/year. They'll pay 20% federal capital gains tax + 3.8% net investment tax + 10.3% CA income tax, or $968 in taxes, and they're left with $2252.
On the other hand if they donate the shares to a charity (or DAF), they get a tax deduction for the appreciated amount ($3220), which can be taken against 35% federal income tax + 10.3% CA income tax = $1459.
So in the scenario where they just sell the shares, the proceeds after taking taxes into account are:
Donor $2252
Charity $0
And in the scenario where they donate the shares, they are:
Donor $1459
Charity $3220
In other words, for an effective cost to the donor of $793, the charity gets $3220.
Isn't that double counting? In the first scenario, $2252 is the amount that they get in their bank account post-tax (marginal). In the second scenario, they get $0 in their bank account post-tax (marginal), and there is a deduction but that occurs on the amount donated so they don't really gain $1459 but rather avoided paying those taxes.
To be clear, this is exactly the same for any gift of appreciated stock -- a DAF is not required to donate appreciated stock in-kind and deduct the full (unrealized) price without paying capital gains. But the vast majority of charities are not equipped to accept in-kind stock donations, which a DAF facilitates by taking stock in-kind and cutting the charity a check.
(Also, DAFs allow claiming the deduction during high-income years and deferring distribution to charities over a longer period of time.)
There is at least one intermediate step: it's not prohibitively expensive to set up a charitable remainder trust. You have more control than with a DAF. But you have a fixed cost to set up the trust and some annual administration and tax compliance costs. It can make sense if you plan to donate more than, say, a million dollars.
I don't have this problem, but getting a "fair" appraisal of your art can be tough. Maybe they auction it off, and the proceeds go to their foundations?
This is more or less what billionaires do to pass their wealth to their children. Here's a recent article that goes into detail about one particular family's setup.
DAFs and GRATs are completely different vehicles. DAFs are beneficially owned by some 501(c)(3) and withdrawals can only go to arms-length, 3rd party 501(c)(3)s, with no reciprocity attached to the gift.
DAFs are a convenient way to gift appreciated stock (which is already a nice tax gift to the charitable wealthy), but fundamentally not a vehicle for passing money to your heirs.
Your article is about GRATs, which should be illegal.
While we're on the topic: you can also leave your estate to a DAF. (If you're married or have kids, probably you should ignore this.)
So that money goes to charity, but what charities? You won't be here, obviously. When you're looking into this, see if your DAF administrator allows a "successor trustee." If not, that institution itself (Schwab, Vanguard, whatever) will disburse it.
If they do, you can pick someone whose values you trust to be the successor & disburse the money. (Probably someone younger than you!) You should ask them, or else they'll get a real surprising phone call right after you die.
On the books since 2011. Upheld in a recent decision of France's supreme court despite what some thought to be quite clearly contrary EU caselaw (which takes precedence over national law, roughly speaking)
https://www.nextinpact.com/article/45613/comment-conseil-det...
From that article: “The internet is generally not anonymous, and if you are breaking Swiss law, a law-abiding company such as ProtonMail can be legally compelled to log your IP address.”
>Because everything in Signal is end-to-end encrypted by default, the broad set of personal information that is typically easy to retrieve in other apps simply doesn’t exist on Signal’s servers.
The E2EE in Signal only protects the actual content of messages. In the case where Signal takes an assertive action, and the users are not paying any attention to their "safety numbers" (probably the most common case) they could in theory get message content with a MITM attack.
With an less assertive action (simply saving the data) Signal could get access to things like contacts and phone numbers.
Tutanota and Protonmail have both been forced in the past to take assertive actions to retain data as a result of legal warrants. Does American law even allow such warrants? If not then perhaps the USA is underrated as a place to base privacy oriented services.
A judge can sign an order commanding a witness or party to preserve documentation and evidence, under penalty of contempt of court. However, there is still a great deal of uncertainty as to what actions the subject of the subpoena must take in order to preserve that evidence. It's pretty clear that you have to disable automated destruction mechanisms, you can't disable any recording functions you may already have, and you can't go and shred relevant papers in your possession; but whether a court can order you to write code or take other burdensome steps in order to record certain electronic records that you didn't record before to assist an ongoing investigation is still a very open question.
Out of curiosity do you if you're within your rights to say "this will cost 'x' amount, we cannot afford it" or say if this is requested we would prefer to dissolve the company?
Basically can the UK government compel you under the threat of criminal prosecution?
I don't know of any instance when someone has threatened to close the company, in the UK (but then again, these notices aren't public knowledge, so just because you don't hear something doesn't mean it doesn't happen).
Noncompliance with a data retention notice lands you in court, but it's a civil matter, not criminal - though if you then ignore an order of the court, you could be facing some jail time and/or a fine for contempt of court. https://www.legislation.gov.uk/ukpga/2016/25/section/95/
Regarding compliance costs, there's a vague provision in the IPA for contribution to your expenditure: https://www.legislation.gov.uk/ukpga/2016/25/section/249
A little more detail, although not much, can be found in the accompanying Communications Data Code of Practice
Sealed sender only means Signal doesn't know who sent a particular message. They have to know who the recipient is so they can deliver it. Like forging the "From:" address on an email. Except in the Signal case the IP address/port of the sender is unique to the user and if the recipient responds then the link between the users is made.
The private contact discovery depends on an Intel SGX hardware enclave on their server. Which is good in this case as it implies more work to bypass it but where is the ultimate trust here? Intel? Did Signal ever get this working?
In general Signal can just see what IP address/port picks up a particular user's pre-keys if they want to know who is talking to who.
> The E2EE in Signal only protects the actual content of messages.
> [By] (simply saving the data) Signal could get access to things like contacts and phone numbers.
And the linked blog posts show that for a few years now they've been working on limiting their own access to this kind of data --- i.e. it's not as simple as just saving it (like e.g. WhatsApp is able to and most likely doing 100% of the time to build social graphs).
Now of course all of those things are likely vulnerable to some attacks, but that's another discussion and it doesn't change the fact that Signal doesn't have immediate easy access in the way you claimed in your original comment. The fact is that there are some barriers and they'd have to put some effort into either disabling these protections or exploit flaws in them to get to the data.
>...it doesn't change the fact that Signal doesn't have immediate easy access in the way you claimed in your original comment.
I did not at all mean to imply that. The assertive action that Signal would have to take might involve actual work. The question is if they could be forced to do that work by the authorities of the country they operate from. I doubt that the amount of work would really factor into the legal stuff.
Signal of course might of already done the work as part of some cooperation with a national signals organization or simply because someone felt bored and contrary, but they could not admit that if they want to preserve the value of the information gathered.
The E2EE encryption part is in the end the only provable aspect of Signal Messenger. The leakage of meta information is inherent when one entity controls all the infrastructure.
Even worse - American laws allow the US government agencies to actually access the servers directly (or even add other servers or routers) in the data centre of the service provider, and the service provider is legally obliged to not tell anyone about it!
Out of interest, is there a place where this practice doesn't exist (genuinely don't know), where service providers are allowed to tell targets they are being subject to a warrant?
Reminder that this does not hold true for Apple's fake "end to end encrypted" iMessage: iCloud Backup, which is not end to end encrypted, uploads all of your iMessages* to Apple each night in a format that Apple can read without you (and turn over to the state upon legal demand such as this).
Note that disabling iCloud Backup won't help you, as it's turned on by default and everyone else you iMessage with will be leaking your conversation plaintext to Apple for you.
Disable iMessage. Use Signal exclusively.
* if you use Messages in iCloud, iCloud Backup instead backs up the cross-device sync key instead of the iMessages themselves, which means Apple gets your iMessages in real time as they sync between your iCloud devices, instead of once per day
It's required. There are statutes tying disclosure of subpoenas to Obstruction charges. This is not a new issue; subpoena secrecy was a thing before there was an Internet.
As an ISP: This is a very boilerplate subpoena. Whether or not the specific FBI agent knows or cares what Signal is, I'm about 99% certain it's just the result of a copy/paste from a template.
Thank you for providing some reality into this farce of a forum. hackernews has completely gone down the toilet! The number of daily topics from NBC/WSJ/etc. is out of control, this might as well be reddit.
Signal is a 501c3 nonprofit- they don't have all that much funding or legal firepower beyond their regular operations. The ACLU also loves them, and getting a letter from the ACLU probably makes matters go away faster then getting a letter from some random lawyer.
What I don't understand about the whole Signal E2EE model is that while your messages themselves may be encrypted, they are still sending push notifications over Apple's servers, which have to go through APNS. Often the entire message contents can be contained in the push notification.
Does anybody know if Apple's notifications are E2EE? I doubt that gov't doesn't have access to the push notifications...
I'm guessing here, but wouldn't they just push the e2ee message through APNS? Then decrypt client side. Or does Apple require plaintext messages for push notifications (that seems bad if they do)?
When you craft a push notification server-side, it contains the payload in plaintext. Now, that is probably encrypted in Apple-land, but my point is that the gov't probably has sunk its teeth into Apple already. So, yea signal's encryption may be open source and proven, but I doubt Apple's doesn't have a backdoor.
Not sure if Signal is doing this, but they could send a notification with title "New message" and encrypted payload. The payload can be processed by a client-side notification extension which decrypts the payload and chooses what notification text the user will see.
I mean Apple themselves is telling devs to not send sensitive data in the actual notification
> [...] never include sensitive data or data that can be retrieved by other means in your payload. Instead, use notifications to alert the user to new information or as a signal that your app has data waiting for it.
Not true. The server can send encrypted (not plaintext) payload. The client can then use a NotificationExtension to decrypt the payload as it comes in.
I would (naively) assume that the notification service sends opaque (encrypted) blobs that are processed (decrypted) by the app before display to the user.
I'm not too familiar with this, but my understanding is that the push notification just wakes up the Signal app, then the Signal app gets the encrypted message (either from Signal's servers or the push notification payload, I'm not sure) and decrypts it client-side and provides the notification text.
I don't know how Signal works but it is possible to send a silent encrypted push notification that the app can decrypt and show as a local notification.
I believe they are encrypted (and decrypted on device by the Signal app).
They recently had to do some rewriting of the code for iOS15 - they share some comments about that here: https://community.signalusers.org/t/beta-feedback-for-the-up...
Hope it helps
I'm actually surprised they didn't use a notification extension before. They're surprisingly great as an API - I used it to dynamically render preview line chart images for a finance app I worked on a few years ago. Just send over the limited line data, render the image, and you're good to go.
I was wondering about the same thing. I think that signal just sends a message to APNS (and Google’s equivalent) that you have something to look at like a new message or whatever. That makes the app wake up and goes to signals servers for the actual content and the app creates the actual notifications on your device.
Even if the push notifications themselves are encrypted, isn't there still the question of whether Apple store the (App x Notification x User/phone number) graph?
This applies on every single app, and is quite irrelevant as you already trust Apple by using their closed source device. If they want your data, they sure get it.
Unless you only contact Signal users who have verified and compiled the client themselves, you put the same kind of trust in Signal, which specify what data is logged (phone numbers are stored hashed for discovery by other users).
The same may or may not be true for Apple (I have no idea) but claiming it is irrelevant as an answer to a question about whether an _Apple_ technology is encrypted, is mind boggling to me.
I mean, there is always a root for the trust. In this case, the root is Apple. Signal is one leaf below. Compromising root compromises everything.
In this case, root being the device and OS, it has unrestricted access to everything happening by your actions. The data you see on your screen is processed by the CPU, developed by Apple, controlled by kernel, coded by Apple. The can access everything they want.
> Often the entire message contents can be contained in the push notification.
Good grief, why would you do that? Just send a notification that data is ready and the when the app wakes, go get the remainder of the data from signal servers.
I actually believe that law enforcement has the legal right to subpoena information, with a judge's consent, while investigating criminal activity. This is exactly the solution to that problem. These platforms should want to know as little about you as possible.
With very limited use, even secret subpoenas can be a good thing, for example in a counterintelligence situation where you don't want to tip your hand to a foreign intelligence service.
...or to your own citizens, who you may be looking at as a domestic threat.
Let's be clear
There is no reason to assume that this type of thing is constrained to "just the type" the government can have their arm forced into admitting to.
That seems to be recent thing (over the last decade or 2) with the US. The Supreme Court of India recently made these observations when the government refused to share certain information with it under the bogey of "national security":
> "... In a democratic country governed by the rule of law, indiscriminate spying on individuals cannot be allowed except with sufficient statutory safeguards, by following the procedure established by law under the Constitution ...
> We had made it clear to the learned Solicitor General on many occasions that we would not push the Respondent-Union of India to provide any information that may affect the national security concerns of the country. However, despite the repeated assurances and opportunities given, ultimately the Respondent-Union of India has placed on record what they call a "limited affidavit", which does not shed any light on their stand or provide any clarity as to the facts of the matter at hand.
> However, this does not mean that the State gets a free pass every time the spectre of "national security" is raised. National security cannot be the bugbear that the judiciary shies away from, by virtue of its mere mentioning. Although this Court should be circumspect in encroaching the domain of national security, no omnibus prohibition can be called for against judicial review.
> The Respondent-Union of India must necessarily plead and prove the facts which indicate that the information sought must be kept secret as their divulgence would affect national security concerns. They must justify the stand that they take before a Court. The mere invocation of national security by the State does not render the Court a mute spectator"
> ... We are not interested in knowing matters related to security or defence. We are only concerned to know whether Govt has used any method other than admissible under law ..."
Yes, although the way around this for law enforcement is to pressure Apple and Google to remove Signal from the App Store/Play respectively (to protect children!) and work on operating system level bypasses of Signal. I am fearing this scenario.
Sideloading on android is quite simple. "download apk" -> "launch apk file" -> "alert gives you a shortcut to settings to allow installing apk from [source]" -> toggle the only switch on that screen -> "launch apk file" now installs it.
You press the only non-"give up" button at each stage and you're done.
Remember that Fortnite succeeded in convincing people to do this by the millions. It's not hard.
Kids hooked on a game vs adults reading a scary message for an app are psychologically very different. Even if fortnite retained millions, how many users did they lose?
>Kids hooked on a game vs adults reading a scary message for an app are psychologically very different.
Somewhat. And sometimes. And when the high-security-site for installing a high-security-tool is asking you to do a thing, it's quite a bit psychologically different than e.g. a reddit post saying "install this apk and approve all prompts: http://www.5z8.info/hack-outlook_w2f7vj_dogfights"
I wonder how far they could go in compelling Signal to push a change that let more info leak for a specific user. I know there have been somewhat similar cases where companies were compelled to add new functionality, logs, etc, to capture info for a specific user.
I'm surprised the FBI has tried to get a custom keyboard into the Play Store yet, or asked Google to add a key logger to the stock one. Sure, the legality is blury at the moment, but it's just a matter of changing some laws and then that becomes legal.
According to Google, Gboard uses Federating Learning to train a model on user data on the local device, so no sensitive data is not sent to the server. Only the gradients are sent and aggregated on the server. https://research.google/pubs/pub47586/
Google has been pretty adamant for years that they don't use or retain your Gboard data, unless you're typing it into search or some Google product that gathers it there. Prediction is supposedly done in-app.
It is perfectly possible to be a grammarly user and not upload everything to it, but if I am running a script or draft of a blog post through it, what harm is there?
I’m curious whether there’s a big overlap between people that the FBI is interested in and people who care about grammar. That’s not a flippant response: I’m genuinely curious if that is part of a shift towards white collar crimes (which could be welcome) and more generally criminal psychology.
I’ve been watching _The Wire_ and Idris Elba’s character sounds like a good overlap, but he feels very unusual (in fiction)--and definitely too careful about OpSec to use that kind of service.
More generally, there’s a lot of anecdotic evidence between problematic behaviour and low-level spelling details: one that is apparently better documented now is a correlation between insurance fraud and whether you capitalise properly the name of your employer on registration form.
Every time I see similar questions, I just wonder why companies such as Signal do not put in place a mechanism to veto changes across countrie.
If they had developers in, say, France, who would need to accept changes made somewhere else then a single country could not force such secret modifications.
Of course an app is actually a binary, so there must be a way to verify that code + compilation = the published binary.
I fully believe that by the end of the decade it will be illegal in some Western countries to go online with a device that doesn't provide a measured boot attestation that it is running a government-approved operating system.
I'm not sure why I was downvoted, but I believe this too.
Same goes for music and video - we won't be legally allowed to hear and watch non-DRM-playable devices as the speakers and displays will be rented not owned by corporations.
We have almost arrived at the Rental Economy, where we dont own anything and are at the behest of corporations. It's modern serfdom
You might have already seen this but in case you havent, there's a famous talk about this called "the coming war on general computation" (https://www.youtube.com/watch?v=HUEvRyemKSg), and much of this has already come true
If there was a heinous criminal and they were using Signal, I wonder if law enforcement could get a warrant to push an update to Signal (or even OS push services) through the Play Store/App Store, just for that user. Wouldn't that work? It's the same legality and hoops to jump through as a wiretap, it seems to me, and it solves law enforcement's problem.
The more I think about it, the more I am convinced they could do this, but they would rather not because it's much harder, since it requires a special warrant with high evidence thresholds for each user. This reinforces the claim that what they really want is an ability to get any messaging data without a warrant, but they don't want to say so.
This is exactly why I don't put 100% trust in these apps that claim e2e encryption. It's trivial to update the app and destroy the contract if the company is willing and cooperating. You need hardware based security to make it truly impossible, like the case of unlocking and iPhone, although I honestly don't know how hardware based security could be leveraged for chat. Maybe in a similar way to how DRM works where the CPU never has access to the data and something else is drawing the pixels on the screen?
I could be wrong but I was under the impression that the way end-to-end encryption worked (like what Signal claims, I thought) was it was physically impossible for them to decrypt (handover decrypted data (aka your messages) to a court of law) because the public/private keys are impossible to crack and also not known by Signal.
It sounds like this isn't the case whatsoever.
I don't really understand modern chat apps that talk about encryption. By no means am I a pro on the subject so I apologize in advance but... if you really don't want ANYBODY EVER snooping on your data network wise (unless they are holding one of the devices and reading the screen after it has been unlocked via passcode/biometric, etc.), can't you just tell your friend a key and exchange it offline and then communicate freely with no middleman? Or even, with a middleman... that is just transporting your data and doesn't know your agreed upon shared secret or keys.
How could a subpoena ever work against this kind of data?
It takes extra effort to design a system with this little amount of data. Note that we only have Signal's word for some of this; they could in fact log every single time that you login, which would make the amount of data sent to the FBI much larger (and could be of importance to the case, for example, if the defendant had a dedicated Signal account for the crime that they only logged into at certain times).
Then there's IPs. If you log IPs along with when someone connects, then an IP can often be tracked to a WiFi router, which then pins your location.
Most E2EE communication protocols will see (and thus potentially log) the time and destination of every message you send. If two people have been accused of conspiring to commit a crime, this could be material in forming the case. They may also store your contact list, but a sufficiently long list of messages sent will practically determine your contact list anyways.
Even just the time of messages could be important; if someone interviewed claimed to be in the shower at a certain time, but there were logs of a message being sent at that time, that's probably enough for an obstruction of justice charge to stick.
> they could in fact log every single time that you login, which would make the amount of data sent to the FBI much larger
I guess you mean each time you connected to their server to retrieve messages? There is no "login" step, Signal doesn't have a log in process. Presumably a typical Signal client calls several times per day.
The record they provided says connection date and despite being supplied in milliseconds since the epoch the value appears to indeed represent a whole day not a specific time. So you're correct Signal could be lying and actually store the exact moment you last connected, for whatever that's worth, and we could not prove they don't have that info.
> Most E2EE communication protocols will see (and thus potentially log) the time and destination of every message you send
Signal doesn't know who sent messages among close friends, and optionally not anybody who sent messages to a sensitive account.
Let's take the example of a message I sent to my friend Chris. I know Chris, I've sent messages to him previously and he knows me, so I'm in his contacts list. As a result by default my Signal client keeps some "Sealed Sender" tokens for Chris. When I send a message to Chris, my client uses a token it learned from a previous conversation with Chris, but Signal doesn't know who I am, just that I have a valid token for sending messages to Chris. So it stores the message, and gives it to Chris. Chris's client can determine that this is a message from me, tialaramex, but the Signal service never knew who sent the message.
If one day Chris hates me and blocks me from sending messages, his client invalidates all Sealed Sender tokens, and begins issuing new ones to his other friends so they can continue to contact him.
> Signal doesn't know who sent messages among close friends, and optionally not anybody who sent messages to a sensitive account.
I am aware of this, and I had split my comment into "extra things Signal could log" and "Extra things most other E2EE services could log" but somehow lost the divider between the two during editing.
I think you misread the post above. They're saying that the govt should be allowed to issue subpoenas, and nothing more. They shouldn't be allowed to mandate backdoors, or hack suspect's machines, etc. And citizens should be free to use cryptography to control their information.
Also, "Impossible" is not the right term. "Extraordinarily expensive" is a better one. And yes, anyone can share public keys with each other offline and have end-to-end encrypted communication without help from a service. But advertising companies and the govt are not incentivized to make that practice convenient, and people typically do what is most convenient.
This doesn't seem that different from subpoening a McDonalds for their security footage, and getting an answer back like "sorry, we've already deleted the footage from that day".
The problem lies not in encryption or key exchange itself. It’s in the fact that to make/use a solution that doesn’t suck in real life, you need a budget comparable to the entire crime you commit. As a developer, you may [help] create few keys and set up the encrypted scene for sending emails or even insecure chat messages, but your gang will fail even at groceries. All these apps are selling turnkey security in app stores, not something brand new.
can't you just tell your friend a key and exchange it offline and then communicate freely with no middleman
Yes, keywords are pubkey, fingerprint, diffie hellman. It’s easy to use, just run:
Now just send .enc files over the wire. It is trivial to decrypt at their side, even my grandma can do that. She usually leaves raw files in her downloads folder though, but it’s easy to remove them via local crontab job.
If neuralink/related tech ever gets to the point of mind-to-mind communication, without a doubt our law enforcement will claim they have the right to subpoena a person's thoughts. We may be setting a precedent now that is more important that we could ever think (pun intended).
That kind of technology would make the issue more obvious, but I already operate under the understanding that people's personal computing devices are effectively extensions of their minds.
No the solution is functioning government. Government can tomorrow make encryption illegal. Why does everyone think the solution is technical when it's a cultural problem? The government has big guns and will use them to enforce the law.
It’s a very small thing, but ~I suspect that~ “Unix timestamp” means they actually send the epoch number, without even bothering to do the conversion to Gregorian data format.
PS: I didn’t notice the attachement. Yup, that’s _exactly_ what they send. Curious whether the suspect could recognise their creation date and know that they are being investigated that way--and whether that exposes Signal.
Exactly, but I fear that government is a step away to ask for ID for any online service. Median age in politics is pretty old and regarding security and experience from the last 20 years suggest they are also easily scared.
While you might forgive someone working for interior ministries, this is an automatism towards fearful stupidity. Benefits don't play are role here, it is exclusively about any potential security benefit, doesn't matter how small.
Sadly signal is based on phone numbers which I dislike, but other messengers should take note here.
Last I checked signal still required a phone number to use so it is an instant deal breaker for a lot of people. I have 3 kids I communicate with but they don’t have a cell number just use wifi when they can. If I could use signal with them I would. Instead I use Wire since it seems secure and doesn’t require a phone number. I can only imagine there are lots of other people with kids in my situation.
Threema is another app I've liked. They have a decent transparency report which shows the limited user data they collect/possess. Link: https://threema.ch/en/transparencyreport
It’s easy so say net win for society is privacy. But it’s important to also acknowledge it does come at a cost — there exists criminal behavior that most reasonable people would agree is bad and should be stopped that may reach a dead end with services like Signal. In formulating your statement that examining criminal behavior is a problem, you are suggesting there shouldn’t be ways to uncover crimes.
The widespread abuse of power in government agencies makes this argument a little naive imo. The vast majority of what they do has very little effect on anyone's safety. I'd rather be able to communicate privately and let people keep selling drugs if they want to.
>the widespread abuse of power[...] The vast majority of what they do has very little effect*
doing a lot of work here. To what degree is that simply anti-governmental sentiment rather than an honest evaluation of the agencies in question?
Say you'd be living in a narco neighborhood in Mexico were cartels regularly shoot civilians up in private wars, have you considered how badly institutions could do in comparison?
Have you considered that the institutions ostensibly serving those civilians might have been infiltrated by the cartels, and that giving them more power might not have the effect you want?
They can pull this information from either the sender or any of the recipients phones. If the government knows the sender, they can arrest them and confiscate the phone.
> What you can't do, is click on a button to spy on people
There's a subpoena in this process that you're glossing over. You can argue that's too easy or too secretive or something, and that's more than fair, but it's not just 'clicking a button'.
Sub poena is basically a rubber stamp after filling out a form. Often done in secret with the barest of oversight. A warrant requires a bit more justification at least.
The obvious counter-argument, unfortunately, is that the government has a clear idea of which requests the FISA court would reject, and therefore never bothers making such requests. That would mean the court is correctly preventing executive overreach and there's nothing to worry about.
This is only true if the companies you're asking for data refuse to provide it _without_ a subpoena. Many companies (let's us AT&T as an example) will provide law enforcement whatever data they ask for without requiring a subpoena.
Bad example: regulate telecommunications companies are required by statute to provide certain information on request. A better example is Amazon with their law enforcement partnerships through Ring.
I assume parent was probably referring more to the subpoena- / warrant- less "creative" solutions that have been discovered, than the typical exhaustion process.
I'm not arguing about the subpoena, I'm arguing against the idea that encrypted solutions are bad.
If you have a subpoena to open a safe, and you realize that you have no tools that are strong enough to open that safe, you don't suddenly blame safes. You don't tell banks they should stop using safes. You don't ask them to create weaker safes robbers can break into.
You try another route.
A subpoena is fair. Asking signal to preemptively not encrypt the data in case we need it later is not.
azinman2 didn't say that there should not be encryption, just that there's a cost, and I think that's a fair statement. Sometimes, 'other methods' are not viable and you're not going to be able to stop the bad guys.
If there is an automated way to get this data, the agencies will do it, and then retroactively generate the paperwork to make it look ok in court if they feel it’s worth their time.
If structurally there is no privacy except some friction with generating paperwork, then we already have solid evidence here in the US in modern times that you defacto have no privacy.
So you want cost the taxpayer significantly with potentially months of unneeded work and expose cops to potentially more danger to ultimately arrive at the same result? How exactly is this better?
Wiretapping is illegal without a warrant. I believe the spirit of the law there implied that wiretapping of [previous, historical conversations] was _always_ illegal, since a wiretap could only be tracking future conversations by its very nature.
The nature of communication has changed, such that all conversations theoretically have a permanent, historical record, despite the intention of those conversations to not have that historical record. It's called "instant messaging", after all, not "perpetual letter writing". It's meant to be an analogue to talking directly with one another.
The path we've gone down where everyone uses a third party to communicate with each other, and that that third party could theoretically record and retain all communications back and forth in perpetuity does not change the _intent_ of the laws as they were written.
The laws were to protect everyone from unreasonable review of their historical actions.
Perhaps you remember that story - I've completely forgotten the source and am having trouble finding it - about the person taken in the night and thrown in front of a judge. He asked what his crimes were, and the judge said "that's what we're here to find out", as they were going to go through everything he'd ever done to find something to charge him with.
"If you give me six lines written by the hand of the most honest of men, I will find something in them which will hang him." -- Attributed (possibly apocryphal) to Cardinal Richelieu (1585-1642).
Ironically, under totalitarian regimes, citizens find themselves faced with the choice between either following the unwritten rules of the regime (which require them to break the laws as written) or following the written laws (which are an impediment to the regime, and thus cause one to be guilty of more nebulous crimes like "sedition").
I don't disagree that overly broad laws are a problem in non-totalitarian countries, but I think that regimes like Stalin's are a special case.
>So you want cost the taxpayer significantly with potentially months of unneeded work and expose cops to potentially more danger to ultimately arrive at the same result? How exactly is this better?
Because my privacy and that of most others who are decent, law-abiding citizens is more important than not making police do their jobs.
How do you think police caught people before apps like Signal? With real police work. Perhaps if they had to spend more time doing that, they wouldn't have time to beat and kill as many unarmed civilians.
The end doesn't justify the means. Police in democratic societies have less power on what they are allowed to do in order to stop crimes, uncover crimes or prosecute criminals. Like requiring a search warrant or how long the police can hold you, interrogate you and so forth. But speech in general has always been a private matter, encryption only reinforces the status quo of society.
What argument do you have that less encryption is the preferred solution?
I have family members that have gone through violent crime that now have PTSD, and due to lack of evidence because of the inability to read chat logs, the perpetrator is free and the case never brought against him.
Meanwhile Encrochat's non-encryption ended up allowing a multinational set of drug cartels to be taken down.
I (obviously) have no idea about the details of that situation, but since a violent crime can't be committed over the internet via a chat app, there ought to be physical evidence of that crime, no?
If there's some sort of conspiracy element to that, I can see how chat logs might be useful.
But attempting to require folks to provide information they don't have (as is the case here) is a fruitless endeavor.
What solution would you suggest? Get rid of encryption? Force providers to collect the contents of their users' computers and phones?
While, as I said, I sympathize with your family members (and you), such an outcome doesn't justify taking away everyone's privacy.
Especially since the vast majority of people are decent, law-abiding folks.
I get that your experiences and the pain they've caused won't allow you to see things differently, but privacy is important, and I for one, won't give mine up without a fight.
I dont wish to come off as unsympathetic because no one should have to experience violent crime. But dont you consider the danger of weakening encryption through law? Facebook and google are giants that have earned all their money because they are able to read everything people say on their platforms in plaintext. Once you weaken encryption for police, you weaken it for everyone with a motive like the police or the state.
No. They’re suggesting that law enforcement should have a valid reason to request someone’s private data such as this process.
You have added that last line yourself, and it appears to suggest that you would prefer all of humanity be constantly surveilled in case it may catch more criminals.
The Fourth Amendment clearly states that law enforcement has to have a subpoena where a judge agrees there's a valid reason to demand private property, with very limited exceptions.
E2E does not require a valid reason. Its only change as far as law enforcement is concerned is to stop monitoring when they do have a valid reason. (Which I think most people feel is as acceptable trade-off.)
> you would prefer all of humanity be constantly surveilled in case it may catch more criminals.
Not only did I not say such a thing (I even said it was easy to argue that encryption is a net win), it’s not something I believe, especially when you put it in such extreme terms. But encryption brings a cost, one that shouldn’t be ignored.
Most people here are taking extreme arguments — assuming everything is about mass surveillance and crimes are more often than not victimless. This ignores the reality that real crimes are regularly happening that most reasonable people would wish to stop, and when you add friction to that, it means there are many cases were justice will not be served.
Those who want to keep the current draconian status quo in place are incentivized to make public any wins, it would justify the existence of the extreme measures. The fact that they haven't boasted about any win is telling.
Yes, and we should also be aware of a rhetorical strategy used by surveillance state apologists where they alternate between "The lack of recent terrorist attacks proves that mass surveillance works, so we should extend it to fight other crimes" and "The recent attack shows that we need more surveillance powers to protect ourselves".
The beauty of the argument is it can never be disproved because all success or failures theoretically need to be kept secret no?
Would YOU be happy spending trillions of dollars with no way to see if it was accomplishing anything, while clear negative effects were also occurring?
This is the cost of abusing the public's willingness to allow certain exceptions to civil liberties. In a society where the public generally trusts the authorities, this problem wouldn't occur. People would almost always be willing to have their communications available for potential judicially-guarded examination, trusting that only justified suspicion of particularly violent crimes will ever be cause for using it.
But when the authorities transgress once too many, the public in general will switch to services that properly defends their privacy.
We can consider this a game-theoretic outcome of abusing the trust of the public. The consequence will eventually be that properly henious criminals will have better tools for not getting caught.
We end up debating trade-offs where people don't agree.
Privacy with end-to-end encryption keeps everyone's communications safe. Criminals, politicians, people working for government contractors, and everyone else. This means criminals can get away with more things. It also means that politicians and surveillance governments have a harder time monitoring regular people or their government challengers.
> In formulating your statement that examining criminal behavior is a problem [...]
Who exactly said this? It's rather the other way around: flagrantly examining and being able to examine non-criminal behaviour at a whim is a problem. The excuse of potentially being able to spot criminal behaviour is not enough.
The GP did: "I actually believe that law enforcement has the legal right to subpoena information, with a judge's consent, while investigating criminal activity. This is exactly the solution to that _problem_." Nothing was said about spotting at large, but the context was subpoenaing information with a judge's consent while investigating criminal activity.
>"I actually believe that law enforcement has the legal right to subpoena information, with a judge's consent, while investigating criminal activity. This is exactly the solution to that _problem_."
Absolutely. The other side of that coin is that people are not required to keep (or in this case, even gather) information in a way that allows the government to obtain it.
I'd also point out that this isn't about information that could prove a crime. It's about the government demanding information from a third party about unknown persons and the contents of their personal effects.
Given that Signal doesn't collect or have access to such information[0]:
"...this subpoena requested a wide variety of information we don’t have, including the target’s name, address, correspondence, contacts, groups, calls."
It's not possible to provide it. Are you claiming that Signal should be required to gather such information solely for the benefit of the police?
As the Fourth Amendment[1] to the US Constitution says, in part:
"...and no Warrants shall issue, but upon probable cause, supported by Oath or affirmation, and particularly describing the place to be searched, and the persons or things to be seized"
And since the subpoena was asking for Signal to identify the subject (their name), such a demand is clearly outside the bounds of the Fourth Amendment.
I'll say it again: Whether a judge (in this case, it was a grand jury and not a judge, but why split hairs?) agrees or not, Signal can't provide information it does not possess.
I suppose a law could be passed requiring them to collect such information as was demanded, but it's hard to see how that would be defensible on any grounds.
So there was no crime before Signal? The "I have nothing to hide so I don't care" argument is so shortsighted. Absolute power corrupts absolutely. Remember this from the Nazi resistance?
First they came for the socialists, and I did not speak out, because I was not a socialist.
Then they came for the trade unionists, and I did not speak out, because I was not a trade unionist.
Then they came for the Jews, and I did not speak out, because I was not a Jew.
Then they came for me, and there was no one left to speak for me.
Now is the time to speak out. By the time you want to protest and push back, it could be too late.
Think about it this way; if the government wants to know something about you, they'll be able to find out. Switching browsers, or search engines, or email providers, or chat apps will not stop them from their goals.
But it can make your life a lot more inconvenient.
The US government is too caught up in prosecuting victimless crimes, bullying defendants into taking plea deals (and forfeiting their right to a fair trial), handing out cruel sentences, and using evidence borne from illegal searches (while lying about it).
Until all of that changes I am not interested in giving them more ammo.
The net benefit to society when government is granted and/or authority is granted broad powers of surveillance is the abuse of that power to serve the desires of those in power rather than society in general.
Your statement is carefully crafted to sidestep this with the wording, "...there exists criminal behavior that most reasonable people would agree is bad and should be stopped that may reach a dead end with services like Signal...", ignoring that the crime of abuse of power is far greater than any crime that could be prevented when it'd granted.
There will always be "some people" that think this way. But more certainly such powers will be abused by those entrusted with them.
>there exists criminal behavior that most reasonable people would agree is bad and should be stopped
Absolutely.
>In formulating your statement that examining criminal behavior is a problem, you are suggesting there shouldn’t be ways to uncover crimes.
I didn't get that at all. Before Signal and other encrypted apps, folks who didn't want to be spied upon would meet in person, in private places or write messages in code.
That didn't stop the police from bringing down many criminals, such as Al Capone, the New York Mafia and many others, did it? Nope, it didn't.
What you seem to be advocating is that everyone's privacy should be forfeited so police can get information without doing, you know, police work.
I'm all for bringing criminals (especially violent ones) to justice. But I'm not willing to give up my privacy so that police can spend their time eating donuts instead of their jobs.
Feel free to disagree, but I'm going to keep using Signal and be glad of it -- not because I'm involved in criminal activity, but because I value my privacy.
You’re attacking a straw man. I never proposed anything other than recognizing the cost of encryption. And if you are to honestly do so, then you also need to recognize things happen now digitally that would have been in person before, which ends up leaving clues like witnesses and DNA.
Apart from just not having encrypted data, the only way to achieve what you're suggesting is with a government backdoor into the encryption.
Any backdoor - any! - will result in your data being exposed, sooner or later. Your Signal messages could then be exposed in a data breach on the dark web for all to see.
It is not worth it to risk everyone's privacy to allow for the chance at easily prosecuting a small number of crimes. Remember - you're not preventing crime this way, just allowing for easy evidence capture. There are viable alternative ways of investigating crimes, as others here have said. There are not viable alternative ways of protecting our data.
Noticed that the last connection time is a date, rounded to the day.
1634169600000 (unix millis)
Thursday, October 14, 2021 12:00:00 AM
Well done. I immediately thought that having a millisecond granularity of last connection time could be used to roughly correlate who contacted whom, depending on what the "connected" event is considered.
I just realized something. One of the only things contained is the account creation date. How hard would it be for the FBI to pull that text you get at that time/date to activate Signal? Not Impossible I would imagine?
Edit: What raised my eyebrow is that the subpoena specifically asks for that. Why?
Shouldn't Signal be required to produce all the encrypted data stored for this user, in case law enforcement are able to get the associated private keys off the suspect's phone?
Signal stores messages on their servers until they're delivered at which point they're purged.
Additionally, Signal's encryption scheme gives their messages the "forward secrecy" property which means that acquiring key material at some point in the future does not allow you to decrypt any previous messages. Any encrypted messages that they could provide would be useless.
For more, check out their really interesting doc on the double ratchet algorithm that they use!:
617 comments
[ 2.3 ms ] story [ 535 ms ] threadWhen encryption and secure messaging is outlawed, only outlaws will have and use it.
Freedom is won in the courts and the legislature, not in the code (although tech is as useful tool for keeping government implantations in check).
(I still use and donate to Signal, but have a healthy understanding of its limits)
Freedom is won through weapons. Encryption is a potent weapon, it can defeat states, militaries. Before computers, it used to be a military tool. It must be democratized, the whole world must use it.
They don't necessarily need to outlaw it. They may just throw up enough hurdles that it doesn't become a major success. Developing a communication system that is secure, featureful and convenient to use for the general population is not a trivial task. A large effort that can be undermined.
E.g. if they only require logging from communication service providers but not from application developers then this would force a decentralized solution. If they lean on payment providers it might get difficult to charge for phone apps or get donations.
The software could continue to legally exist but see little adoption. Which is enough to enable surveillance.
https://en.wikipedia.org/wiki/NSA_cryptography
It's probably fine here, but if you store it in binary, you should probably parse it into something human-readable.
It's the final dash on the icing of "politely F yourself". Compliant and accurate but "let me burn up a little bit _your_ time" (pun).
Made me smile.
https://signal.org/donate/
Most "donate" pages do not allow for "donor-advised funds (DAF)." They assume you're giving it with your before-tax money and presumably taking a tax deduction for it.
In a DAF, which your financial institution surely offers, you can donate appreciated assets, e.g. your FAANG stock, and take the entire amount as a tax deduction. So if your 10 shares of Facebook (excuse me, "Meta") stock are at 322, you can take a deduction of $32,200 this year.
What's the catch? That money's gone, and you can't get it back. You can only "advise" your DAF to give it to a 501(c)(3) organization, which Signal is. There are no time limits.
The good part, though, is you can probably have your DAF give the money anonymously, so the charity can't bug you every time they're having a fund drive.
Like, let's say your intent is to donate $10k to some charity, out of the goodness of your heart and/or as a tax write off. You don't have that in cash, but do in stock.
You could liquidate $10k of stock, pay capital gains on it (if it appreciated since acquisition), then donate it. So you're out the capital gains tax.
The method you describe seems more efficient, since you don't need to sell; you simply transfer ownership of the asset.
Or is there still capital gains to be paid?
I wonder if billionaires are setting up charities as trusts for their kids, then "donating their shares to charity?"
Billionaires have access to much fancier schemes than this, and I won't even attempt to describe all those. But yeah, I imagine "donating their shares without capital gains taxes" figures into them.
I just noticed you said "trusts for their kids" -- that's something different. If the children can access it, it's not a DAF. But trusts are much more complicated, and someone who understands them (which I don't) can hold forth here.
In your example above, let's say the person purchased those 10 Meta shares for $38 each at the IPO and they're worth $322 each now. That's $3220 in proceeds and a $2840 capital gain.
The taxes on this depend on income level and state of residence, but let's say they're in CA making $300K/year. They'll pay 20% federal capital gains tax + 3.8% net investment tax + 10.3% CA income tax, or $968 in taxes, and they're left with $2252.
On the other hand if they donate the shares to a charity (or DAF), they get a tax deduction for the appreciated amount ($3220), which can be taken against 35% federal income tax + 10.3% CA income tax = $1459.
So in the scenario where they just sell the shares, the proceeds after taking taxes into account are:
And in the scenario where they donate the shares, they are: In other words, for an effective cost to the donor of $793, the charity gets $3220.If you donate to a DAF, it's 100% gone to charity, *someday."
(Also, DAFs allow claiming the deduction during high-income years and deferring distribution to charities over a longer period of time.)
https://archive.md/yN7M7
https://www.bloomberg.com/features/how-billionaires-pass-wea...
DAFs are a convenient way to gift appreciated stock (which is already a nice tax gift to the charitable wealthy), but fundamentally not a vehicle for passing money to your heirs.
Your article is about GRATs, which should be illegal.
Also, the annual stock deduction limit is capped at 30% or so of income.
So that money goes to charity, but what charities? You won't be here, obviously. When you're looking into this, see if your DAF administrator allows a "successor trustee." If not, that institution itself (Schwab, Vanguard, whatever) will disburse it.
If they do, you can pick someone whose values you trust to be the successor & disburse the money. (Probably someone younger than you!) You should ask them, or else they'll get a real surprising phone call right after you die.
https://www.legifrance.gouv.fr/jorf/id/JORFTEXT000023646013
On the books since 2011. Upheld in a recent decision of France's supreme court despite what some thought to be quite clearly contrary EU caselaw (which takes precedence over national law, roughly speaking) https://www.nextinpact.com/article/45613/comment-conseil-det...
https://threatpost.com/protonmail-log-ip-address-french-acti...
From that article: “The internet is generally not anonymous, and if you are breaking Swiss law, a law-abiding company such as ProtonMail can be legally compelled to log your IP address.”
They must keep this data hot because they can send "this specific person in your phonebook just installed signal" messages.
The E2EE in Signal only protects the actual content of messages. In the case where Signal takes an assertive action, and the users are not paying any attention to their "safety numbers" (probably the most common case) they could in theory get message content with a MITM attack.
With an less assertive action (simply saving the data) Signal could get access to things like contacts and phone numbers.
Tutanota and Protonmail have both been forced in the past to take assertive actions to retain data as a result of legal warrants. Does American law even allow such warrants? If not then perhaps the USA is underrated as a place to base privacy oriented services.
https://www.legislation.gov.uk/ukpga/2016/25/section/87/enac...
Ctrl+f for "generation"
Basically can the UK government compel you under the threat of criminal prosecution?
Noncompliance with a data retention notice lands you in court, but it's a civil matter, not criminal - though if you then ignore an order of the court, you could be facing some jail time and/or a fine for contempt of court. https://www.legislation.gov.uk/ukpga/2016/25/section/95/
Regarding compliance costs, there's a vague provision in the IPA for contribution to your expenditure: https://www.legislation.gov.uk/ukpga/2016/25/section/249 A little more detail, although not much, can be found in the accompanying Communications Data Code of Practice
I am a lawyer, but I am not YOUR lawyer.
They use sealed sender: https://signal.org/blog/sealed-sender/
Private contact discovery: https://signal.org/blog/private-contact-discovery/
And a "Private Group System" which is supposed to keep group membership information from the server: https://signal.org/blog/signal-private-group-system/
Though of course they could still push malicious updates.
The private contact discovery depends on an Intel SGX hardware enclave on their server. Which is good in this case as it implies more work to bypass it but where is the ultimate trust here? Intel? Did Signal ever get this working?
In general Signal can just see what IP address/port picks up a particular user's pre-keys if they want to know who is talking to who.
> The E2EE in Signal only protects the actual content of messages.
> [By] (simply saving the data) Signal could get access to things like contacts and phone numbers.
And the linked blog posts show that for a few years now they've been working on limiting their own access to this kind of data --- i.e. it's not as simple as just saving it (like e.g. WhatsApp is able to and most likely doing 100% of the time to build social graphs).
Now of course all of those things are likely vulnerable to some attacks, but that's another discussion and it doesn't change the fact that Signal doesn't have immediate easy access in the way you claimed in your original comment. The fact is that there are some barriers and they'd have to put some effort into either disabling these protections or exploit flaws in them to get to the data.
I did not at all mean to imply that. The assertive action that Signal would have to take might involve actual work. The question is if they could be forced to do that work by the authorities of the country they operate from. I doubt that the amount of work would really factor into the legal stuff.
Signal of course might of already done the work as part of some cooperation with a national signals organization or simply because someone felt bored and contrary, but they could not admit that if they want to preserve the value of the information gathered.
The E2EE encryption part is in the end the only provable aspect of Signal Messenger. The leakage of meta information is inherent when one entity controls all the infrastructure.
Even worse - American laws allow the US government agencies to actually access the servers directly (or even add other servers or routers) in the data centre of the service provider, and the service provider is legally obliged to not tell anyone about it!
Note that disabling iCloud Backup won't help you, as it's turned on by default and everyone else you iMessage with will be leaking your conversation plaintext to Apple for you.
Disable iMessage. Use Signal exclusively.
* if you use Messages in iCloud, iCloud Backup instead backs up the cross-device sync key instead of the iMessages themselves, which means Apple gets your iMessages in real time as they sync between your iCloud devices, instead of once per day
https://mobile.reuters.com/article/amp/idUSKBN1ZK1CT is why fake pro-privacy Apple will never be able to run a story like Signal has here today.
I thought the ACLU was more of a protection against smaller entities who didn't have funding/legal firepower?
Does anybody know if Apple's notifications are E2EE? I doubt that gov't doesn't have access to the push notifications...
> [...] never include sensitive data or data that can be retrieved by other means in your payload. Instead, use notifications to alert the user to new information or as a signal that your app has data waiting for it.
https://developer.apple.com/library/archive/documentation/Ne...
https://developer.apple.com/documentation/usernotifications/...
Edit: wow people were fast to reply…
The same may or may not be true for Apple (I have no idea) but claiming it is irrelevant as an answer to a question about whether an _Apple_ technology is encrypted, is mind boggling to me.
In this case, root being the device and OS, it has unrestricted access to everything happening by your actions. The data you see on your screen is processed by the CPU, developed by Apple, controlled by kernel, coded by Apple. The can access everything they want.
Good grief, why would you do that? Just send a notification that data is ready and the when the app wakes, go get the remainder of the data from signal servers.
https://developer.apple.com/documentation/usernotifications/...
I actually believe that law enforcement has the legal right to subpoena information, with a judge's consent, while investigating criminal activity. This is exactly the solution to that problem. These platforms should want to know as little about you as possible.
The problem is abuse of that system.
Let's be clear There is no reason to assume that this type of thing is constrained to "just the type" the government can have their arm forced into admitting to.
> "... In a democratic country governed by the rule of law, indiscriminate spying on individuals cannot be allowed except with sufficient statutory safeguards, by following the procedure established by law under the Constitution ...
> We had made it clear to the learned Solicitor General on many occasions that we would not push the Respondent-Union of India to provide any information that may affect the national security concerns of the country. However, despite the repeated assurances and opportunities given, ultimately the Respondent-Union of India has placed on record what they call a "limited affidavit", which does not shed any light on their stand or provide any clarity as to the facts of the matter at hand.
> However, this does not mean that the State gets a free pass every time the spectre of "national security" is raised. National security cannot be the bugbear that the judiciary shies away from, by virtue of its mere mentioning. Although this Court should be circumspect in encroaching the domain of national security, no omnibus prohibition can be called for against judicial review.
> The Respondent-Union of India must necessarily plead and prove the facts which indicate that the information sought must be kept secret as their divulgence would affect national security concerns. They must justify the stand that they take before a Court. The mere invocation of national security by the State does not render the Court a mute spectator"
> ... We are not interested in knowing matters related to security or defence. We are only concerned to know whether Govt has used any method other than admissible under law ..."
Source: Supreme Court Constitutes Independent Expert Committee To Probe Pegasus Snooping Allegations - https://web.archive.org/web/20211029130706/https://www.livel...
For iOS users, that will be a death knell.
You press the only non-"give up" button at each stage and you're done.
Remember that Fortnite succeeded in convincing people to do this by the millions. It's not hard.
Somewhat. And sometimes. And when the high-security-site for installing a high-security-tool is asking you to do a thing, it's quite a bit psychologically different than e.g. a reddit post saying "install this apk and approve all prompts: http://www.5z8.info/hack-outlook_w2f7vj_dogfights"
Come to think of it, that'd be the perfect place to go to demand a wiretap - at least one such popular "LY" service already exists.
I'm still shaking my head at what many regular users will agree to..
I’ve been watching _The Wire_ and Idris Elba’s character sounds like a good overlap, but he feels very unusual (in fiction)--and definitely too careful about OpSec to use that kind of service.
More generally, there’s a lot of anecdotic evidence between problematic behaviour and low-level spelling details: one that is apparently better documented now is a correlation between insurance fraud and whether you capitalise properly the name of your employer on registration form.
If they had developers in, say, France, who would need to accept changes made somewhere else then a single country could not force such secret modifications.
Of course an app is actually a binary, so there must be a way to verify that code + compilation = the published binary.
The Internet may interpret censorship as damage and route around it, but spy agencies interpret laws as inconveniences and ignore them.
As access closes in one place (i.e application layer), they will just get closer to the source (i.e operating system or supply chain)
Same goes for music and video - we won't be legally allowed to hear and watch non-DRM-playable devices as the speakers and displays will be rented not owned by corporations.
We have almost arrived at the Rental Economy, where we dont own anything and are at the behest of corporations. It's modern serfdom
The more I think about it, the more I am convinced they could do this, but they would rather not because it's much harder, since it requires a special warrant with high evidence thresholds for each user. This reinforces the claim that what they really want is an ability to get any messaging data without a warrant, but they don't want to say so.
I could be wrong but I was under the impression that the way end-to-end encryption worked (like what Signal claims, I thought) was it was physically impossible for them to decrypt (handover decrypted data (aka your messages) to a court of law) because the public/private keys are impossible to crack and also not known by Signal.
It sounds like this isn't the case whatsoever.
I don't really understand modern chat apps that talk about encryption. By no means am I a pro on the subject so I apologize in advance but... if you really don't want ANYBODY EVER snooping on your data network wise (unless they are holding one of the devices and reading the screen after it has been unlocked via passcode/biometric, etc.), can't you just tell your friend a key and exchange it offline and then communicate freely with no middleman? Or even, with a middleman... that is just transporting your data and doesn't know your agreed upon shared secret or keys.
How could a subpoena ever work against this kind of data?
Then there's IPs. If you log IPs along with when someone connects, then an IP can often be tracked to a WiFi router, which then pins your location.
Most E2EE communication protocols will see (and thus potentially log) the time and destination of every message you send. If two people have been accused of conspiring to commit a crime, this could be material in forming the case. They may also store your contact list, but a sufficiently long list of messages sent will practically determine your contact list anyways.
Even just the time of messages could be important; if someone interviewed claimed to be in the shower at a certain time, but there were logs of a message being sent at that time, that's probably enough for an obstruction of justice charge to stick.
I guess you mean each time you connected to their server to retrieve messages? There is no "login" step, Signal doesn't have a log in process. Presumably a typical Signal client calls several times per day.
The record they provided says connection date and despite being supplied in milliseconds since the epoch the value appears to indeed represent a whole day not a specific time. So you're correct Signal could be lying and actually store the exact moment you last connected, for whatever that's worth, and we could not prove they don't have that info.
> Most E2EE communication protocols will see (and thus potentially log) the time and destination of every message you send
Signal doesn't know who sent messages among close friends, and optionally not anybody who sent messages to a sensitive account.
Let's take the example of a message I sent to my friend Chris. I know Chris, I've sent messages to him previously and he knows me, so I'm in his contacts list. As a result by default my Signal client keeps some "Sealed Sender" tokens for Chris. When I send a message to Chris, my client uses a token it learned from a previous conversation with Chris, but Signal doesn't know who I am, just that I have a valid token for sending messages to Chris. So it stores the message, and gives it to Chris. Chris's client can determine that this is a message from me, tialaramex, but the Signal service never knew who sent the message.
If one day Chris hates me and blocks me from sending messages, his client invalidates all Sealed Sender tokens, and begins issuing new ones to his other friends so they can continue to contact him.
I am aware of this, and I had split my comment into "extra things Signal could log" and "Extra things most other E2EE services could log" but somehow lost the divider between the two during editing.
Also, "Impossible" is not the right term. "Extraordinarily expensive" is a better one. And yes, anyone can share public keys with each other offline and have end-to-end encrypted communication without help from a service. But advertising companies and the govt are not incentivized to make that practice convenient, and people typically do what is most convenient.
Issue a subpoenas for something they know they can't architecturally fulfill? I would imagine the government will get mad and shut them down, no?
can't you just tell your friend a key and exchange it offline and then communicate freely with no middleman
Yes, keywords are pubkey, fingerprint, diffie hellman. It’s easy to use, just run:
to generate a key pair, then export pubkey via: Once you have you exchanged pubkeys and checked fingerprints offline, simply create a new secret key: and encrypt that new key and also your message with it: Now just send .enc files over the wire. It is trivial to decrypt at their side, even my grandma can do that. She usually leaves raw files in her downloads folder though, but it’s easy to remove them via local crontab job.I read a good short story on the subject recently, "The Truth of Fact, the Truth of Feeling": https://en.wikipedia.org/wiki/The_Truth_of_Fact,_the_Truth_o...
PS: I didn’t notice the attachement. Yup, that’s _exactly_ what they send. Curious whether the suspect could recognise their creation date and know that they are being investigated that way--and whether that exposes Signal.
While you might forgive someone working for interior ministries, this is an automatism towards fearful stupidity. Benefits don't play are role here, it is exclusively about any potential security benefit, doesn't matter how small.
Sadly signal is based on phone numbers which I dislike, but other messengers should take note here.
doing a lot of work here. To what degree is that simply anti-governmental sentiment rather than an honest evaluation of the agencies in question?
Say you'd be living in a narco neighborhood in Mexico were cartels regularly shoot civilians up in private wars, have you considered how badly institutions could do in comparison?
But I’d say that we should work to make it impossible for mass surveillance to exist, full stop.
Police should have to do real actual detective work to implicate people in a crime.
That's reasonable.
https://www.lawtechnologytoday.org/2019/08/can-police-force-...
And that's a good thing.
Now, cops and politicians want to solve all the problems from their desk.
No, sorry, my freedom is not to be sold for their convenience.
You want to catch a bad guy, you get a trained investigation team that follows people, that wires their house, that interrogates neighborhood, etc.
Is it more work ? Yes. Is it more dangerous ? Hell, yes.
But don't say you can't catch criminals because of Signal. What you can't do, is click on a button to spy on people. It's a good thing.
This mantra is just an excuse to chew off chunks of democracy.
There's a subpoena in this process that you're glossing over. You can argue that's too easy or too secretive or something, and that's more than fair, but it's not just 'clicking a button'.
Do you have a source for that?
If you have a subpoena to open a safe, and you realize that you have no tools that are strong enough to open that safe, you don't suddenly blame safes. You don't tell banks they should stop using safes. You don't ask them to create weaker safes robbers can break into.
You try another route.
A subpoena is fair. Asking signal to preemptively not encrypt the data in case we need it later is not.
Those are the tip of the iceberg.
If there is an automated way to get this data, the agencies will do it, and then retroactively generate the paperwork to make it look ok in court if they feel it’s worth their time.
If structurally there is no privacy except some friction with generating paperwork, then we already have solid evidence here in the US in modern times that you defacto have no privacy.
[citation needed]
Further:
Wiretapping is illegal without a warrant. I believe the spirit of the law there implied that wiretapping of [previous, historical conversations] was _always_ illegal, since a wiretap could only be tracking future conversations by its very nature.
The nature of communication has changed, such that all conversations theoretically have a permanent, historical record, despite the intention of those conversations to not have that historical record. It's called "instant messaging", after all, not "perpetual letter writing". It's meant to be an analogue to talking directly with one another.
The path we've gone down where everyone uses a third party to communicate with each other, and that that third party could theoretically record and retain all communications back and forth in perpetuity does not change the _intent_ of the laws as they were written.
The laws were to protect everyone from unreasonable review of their historical actions.
Perhaps you remember that story - I've completely forgotten the source and am having trouble finding it - about the person taken in the night and thrown in front of a judge. He asked what his crimes were, and the judge said "that's what we're here to find out", as they were going to go through everything he'd ever done to find something to charge him with.
edit: another instance would be Lavrentiy Beria, a police chief under Stalin (https://www.oxfordeagle.com/2018/05/09/show-me-the-man-and-i...)
"Show me the man, I'll show you the crime."
That's nothing new, either.
"If you give me six lines written by the hand of the most honest of men, I will find something in them which will hang him." -- Attributed (possibly apocryphal) to Cardinal Richelieu (1585-1642).
I don't disagree that overly broad laws are a problem in non-totalitarian countries, but I think that regimes like Stalin's are a special case.
Because my privacy and that of most others who are decent, law-abiding citizens is more important than not making police do their jobs.
How do you think police caught people before apps like Signal? With real police work. Perhaps if they had to spend more time doing that, they wouldn't have time to beat and kill as many unarmed civilians.
If you have someone being blackmailed or defrauded online, what "field work" are you suggesting here?
> "cops [...] want to solve all the problems from their desk."
This literally couldn't be more wrong.
What argument do you have that less encryption is the preferred solution?
Meanwhile Encrochat's non-encryption ended up allowing a multinational set of drug cartels to be taken down.
It's not difficult to come up with such examples.
I (obviously) have no idea about the details of that situation, but since a violent crime can't be committed over the internet via a chat app, there ought to be physical evidence of that crime, no?
If there's some sort of conspiracy element to that, I can see how chat logs might be useful.
But attempting to require folks to provide information they don't have (as is the case here) is a fruitless endeavor.
What solution would you suggest? Get rid of encryption? Force providers to collect the contents of their users' computers and phones?
While, as I said, I sympathize with your family members (and you), such an outcome doesn't justify taking away everyone's privacy.
Especially since the vast majority of people are decent, law-abiding folks.
I get that your experiences and the pain they've caused won't allow you to see things differently, but privacy is important, and I for one, won't give mine up without a fight.
You have added that last line yourself, and it appears to suggest that you would prefer all of humanity be constantly surveilled in case it may catch more criminals.
E2E does not require a valid reason. Its only change as far as law enforcement is concerned is to stop monitoring when they do have a valid reason. (Which I think most people feel is as acceptable trade-off.)
Not only did I not say such a thing (I even said it was easy to argue that encryption is a net win), it’s not something I believe, especially when you put it in such extreme terms. But encryption brings a cost, one that shouldn’t be ignored.
Most people here are taking extreme arguments — assuming everything is about mass surveillance and crimes are more often than not victimless. This ignores the reality that real crimes are regularly happening that most reasonable people would wish to stop, and when you add friction to that, it means there are many cases were justice will not be served.
The privacy/security trade-off is vastly overstated.
Regardless, there are far more ordinary crimes being committed than terrorism.
Would YOU be happy spending trillions of dollars with no way to see if it was accomplishing anything, while clear negative effects were also occurring?
But when the authorities transgress once too many, the public in general will switch to services that properly defends their privacy.
We can consider this a game-theoretic outcome of abusing the trust of the public. The consequence will eventually be that properly henious criminals will have better tools for not getting caught.
Privacy with end-to-end encryption keeps everyone's communications safe. Criminals, politicians, people working for government contractors, and everyone else. This means criminals can get away with more things. It also means that politicians and surveillance governments have a harder time monitoring regular people or their government challengers.
Who exactly said this? It's rather the other way around: flagrantly examining and being able to examine non-criminal behaviour at a whim is a problem. The excuse of potentially being able to spot criminal behaviour is not enough.
Absolutely. The other side of that coin is that people are not required to keep (or in this case, even gather) information in a way that allows the government to obtain it.
I'd also point out that this isn't about information that could prove a crime. It's about the government demanding information from a third party about unknown persons and the contents of their personal effects.
Given that Signal doesn't collect or have access to such information[0]:
"...this subpoena requested a wide variety of information we don’t have, including the target’s name, address, correspondence, contacts, groups, calls."
It's not possible to provide it. Are you claiming that Signal should be required to gather such information solely for the benefit of the police?
As the Fourth Amendment[1] to the US Constitution says, in part:
"...and no Warrants shall issue, but upon probable cause, supported by Oath or affirmation, and particularly describing the place to be searched, and the persons or things to be seized"
And since the subpoena was asking for Signal to identify the subject (their name), such a demand is clearly outside the bounds of the Fourth Amendment.
I'll say it again: Whether a judge (in this case, it was a grand jury and not a judge, but why split hairs?) agrees or not, Signal can't provide information it does not possess.
I suppose a law could be passed requiring them to collect such information as was demanded, but it's hard to see how that would be defensible on any grounds.
[0] https://signal.org/bigbrother/cd-california-grand-jury/
[1] https://en.wikipedia.org/wiki/Fourth_Amendment_to_the_United...
First they came for the socialists, and I did not speak out, because I was not a socialist. Then they came for the trade unionists, and I did not speak out, because I was not a trade unionist. Then they came for the Jews, and I did not speak out, because I was not a Jew. Then they came for me, and there was no one left to speak for me.
Now is the time to speak out. By the time you want to protest and push back, it could be too late.
But it can make your life a lot more inconvenient.
Until all of that changes I am not interested in giving them more ammo.
Well, the other bar is "the justice system follows its own rules." That's reasonable enough to ask, isn't it?
Your statement is carefully crafted to sidestep this with the wording, "...there exists criminal behavior that most reasonable people would agree is bad and should be stopped that may reach a dead end with services like Signal...", ignoring that the crime of abuse of power is far greater than any crime that could be prevented when it'd granted.
There will always be "some people" that think this way. But more certainly such powers will be abused by those entrusted with them.
Absolutely.
>In formulating your statement that examining criminal behavior is a problem, you are suggesting there shouldn’t be ways to uncover crimes.
I didn't get that at all. Before Signal and other encrypted apps, folks who didn't want to be spied upon would meet in person, in private places or write messages in code.
That didn't stop the police from bringing down many criminals, such as Al Capone, the New York Mafia and many others, did it? Nope, it didn't.
What you seem to be advocating is that everyone's privacy should be forfeited so police can get information without doing, you know, police work.
I'm all for bringing criminals (especially violent ones) to justice. But I'm not willing to give up my privacy so that police can spend their time eating donuts instead of their jobs.
Feel free to disagree, but I'm going to keep using Signal and be glad of it -- not because I'm involved in criminal activity, but because I value my privacy.
The world isn’t black and white.
I assume you're referring to this sentence in my comment:
Note that I said seem. Which, in this context, means that's what I understood you to be saying. Thank you for clarifying.What's more, I'm not attacking anything or anyone. Rather, I'm expositing my views WRT encrypted communications and police work.
That you interpreted the expression of my views as an attack says more about you than about me, IMHO.
[1] https://en.wikipedia.org/wiki/Straw_man
What is the "strawman" you appear to believe that I've set up and then attempted to knock down?
Any backdoor - any! - will result in your data being exposed, sooner or later. Your Signal messages could then be exposed in a data breach on the dark web for all to see.
It is not worth it to risk everyone's privacy to allow for the chance at easily prosecuting a small number of crimes. Remember - you're not preventing crime this way, just allowing for easy evidence capture. There are viable alternative ways of investigating crimes, as others here have said. There are not viable alternative ways of protecting our data.
> Account created: 1606866784432 (unix millis)
This response of the user information they have is hilarious.
Thu 14 Oct 2021 12:00:00 AM UTC
Do they round?
They've only gotten used to going after the intermediary, and it feels uncomfortable for them to have this power removed and reset back to the mean.
https://www.aclu.org/ https://signal.org/donate/
Nothing is free, support these folks.
https://www.aclu.org/ https://signal.org/
Nothing is free, support these folks.
Edit: What raised my eyebrow is that the subpoena specifically asks for that. Why?
Additionally, Signal's encryption scheme gives their messages the "forward secrecy" property which means that acquiring key material at some point in the future does not allow you to decrypt any previous messages. Any encrypted messages that they could provide would be useless.
For more, check out their really interesting doc on the double ratchet algorithm that they use!:
https://signal.org/docs/specifications/doubleratchet/