So basically their marketing-department is abusing a security term in order to sound good, as opposed to a software flaw.
They're claiming "end to end" encryption, which usually implies the service is unable to spy on individual users that are communicating to one-another over an individualized channel.
However in this case there are no other users, and their server is one of the "ends" doing the communicating, which is... perhaps not a literal contradiction in terms, but certainly breaking the spirit of the phrase.
This is an incredibly common misuse of the term e2ee. I think at this point we need a new word because you have a coin flip's chance of actually getting what you think when a company describes their product this way.
End-to-end encryption doesn't mean anything where it is semi-validly used. It's used on phones, where you as a user (or company) don't control what code executes. For example, WhatsApp was end-to-end encrypted. Well, it doesn't actually provide security because with either physical access to the phone or if you have if you can use the app store to "upgrade" the app, you can upload code to the phone. You can upload an apk that replaces the WhatsApp app. It even still uploads the messages to a central server so you can get those messages from Meta, then get the key from the phone some time later or earlier and use the key to decrypt it when the message is already erased from the phone.
(aside from the fact that people don't seem to know/remember WhatsApp backs up to google drive)
Code that then gets access to the end-to-end encryption keys ... so you're not safe from state actors, you're not safe from police, you're not safe from the authors of the code and you're not safe from anyone who has physical access to your phone.
There was a discussion here on hn about OpenAI and it's privacy. Same confusion about e2ee. Users thinking e2ee is possible when you chat with an ai agent.
>Users thinking e2ee is possible when you chat with an ai agent.
It shouldn't be any harder than e2ee chatting with any other user. It's just instead of the other end chatting using a keyboard as an input they chat using a language model to type the messages. Of course like any other e2ee solution, the person you are talking to also has access to your messages as that's the whole point, being able to talk to them.
I do not think this matches anyones' mental model of what "end-to-end encrypted" for a conversation between me and what is ostensibly my own computer should look like.
If you promise end-to-end encryption, and later it turns out your employees have been reading my chat transcripts...
I'm not sure how you can call chatgpt "ostensibly my own computer" when it's primarily a website.
And honestly, E2EE's strict definition (messages between user 1 and user 2 cannot be decrypted by message platform)... Is unambiguously possible for chatGPT. It's just utterly pointless when user2 happens to also be the message platform.
If you message support for $chat_platform (if there is such a thing) do you expect them to be unable to read the messages?
It's still a disingenuous use of the term. And, if TFA is anything like multiple other providers, it's going to be "oh, the video is E2EE. But the 5fps ,non-sensitive' 512*512px preview isn't."
> it's primarily a website … unambiguously possible[sic] for chatGPT … happens to also be the message platform
I assume you mean impossible, and in either case that’s not quite accurate. The “end” is a specific AI model you wished to communicate with, not the platform. You’re suggesting they are one and the same, but they are not and Google proves that with their own secure LLM offering.
But I’m 100% with you on it being a disingenuous use.
No, No typo- the problem with ChatGPT is the third party that would would be Attesting that's how it works, is just the 2nd party.
I'm not familiar with the referenced Google secure LLM, but offhand- if it's TEE based- Google would be publishing auditable/signed images and Intel/AMD would be the third party Attesting that's whats actually running. TEEs are way out of my expertise though, and there's a ton of places and ways for it to break down.
> And honestly, E2EE's strict definition (messages between user 1 and user 2 cannot be decrypted by message platform)... Is unambiguously possible for chatGPT. It's just utterly pointless when user2 happens to also be the message platform.
This is basically the whole thrust of Apple's Private Cloud Compute architecture. It is possible to build a system that prevents user2 from reading the chats, but it's not clear that most companies want to work within those restrictions.
> If you message support for $chat_platform (if there is such a thing) do you expect them to be unable to read the messages?
If they marketed it as end-to-end encrypted? 100%, unambiguously, yes. And certainly not without I, as the user, granting them access permissions to do so.
If you have an E2EE chat with McDonalds, you shouldn't be surprised that McDonalds employees can read the messages you've sent that account. When messaging accounts controlled by businesses then the business can see those messages.
This is why I specified "mental model". Interacting with ChatGPT is not marketed as "sending messages to OpenAI the company". It is implied to be "sending messages to my personal AI assistant".
Yeah of course, technically that is true. Still when talking about e2ee in any context it implies to the non technical user: The company providing the service cannot read what I am writing.
That's not given in any of those examples. In the case of ChatGPT and this toilet sensor e2ee is almost equivalent to 'we use https'. But nowadays everybody uses https, so it does not sound as good in marketing.
Yes but National Security Letters make that pointless. You can't encrypt away a legal obligation. The point of e2ee is that a provider can say to the feds "this is all the information we have", and removing the e2ee would be noticed by security researchers.
If the provider controls one of the ends then the feds can instruct them to tap that end and nobody is any the wiser.
The best you can do is either to run the inference in an outside jurisdiction (hard for large scale AI), or to attempt a warrant canary.
> Yes but National Security Letters make that pointless
It seems ridiculous to use the term "national security letter" as opposed to "subpoena" in this context, there is no relevant distinction between the two when it comes to this subject. A pointless distraction.
> You can't encrypt away a legal obligation.
Of course you can't. But a subpoena (or a NSL, which is a subpoena) can only mandate you to provide information which you have within your control. It can not mandate you to procure information which you do not have within your control.
If you implement e2ee, customer chats are not within your control. There is no way to breach that with a subpoena. A subpoena can not force you to implement a backdoor or disable e2ee.
I believe we are in agreement. If you are a communication platform that implements e2ee then you provide the guarantee to users, backed by security researchers, that the government can't read their communications by getting a subpoena from the communication platform.
The problem with AI platforms is that they are also a party to the communication, therefore they can indeed be forced to reveal chats, and therefore it's not e2ee because defining e2ee that way would render the term without distinction.
They once shipped a backdoor in their macOS app. It was noticed and called out and they refused to remove it. It took Apple blacklisting it for Zoom to finally take action.
I would say Telegram is communicating their level of encryption pretty good ("client-to-client" and "client-to-server" is a good way to avoid the ambiguity of e2e).
The problem is that you have to trust that they'll stay that way, and we have no way of proving that the app that runs on your phone comes from the same source that they publish.
It's not incredibly common, there's sure a lot of companies that try to misuse it, but the average person (even non technical) still interprets it in the correct way
I think part of the problem is that prior to WhatsApp's E2EE implementation in like 2014, TLS was very often called "End to End Encryption" as the ends were Client and Server/Service Provider. It got redefined and now the new usage is way more popular than the old one.
I can't blame most people for calling TLS "E2EE", even some folks in industry, but it's not great for a company to advertise that you offer X if the meaning of X has shifted so drastically in the last decade.
I’m pushing back on that one. I’ve been running websites since the ‘90s, and I’ve never heard E2EE used that way until very recently by vendors who, bluntly, want to lie about it.
It was pretty common to call client-side encryption/SSL "end to end encryption" among network engineers who were analyzing data flowing through their networks[0] as well as those who were implementing SSL/TLS into their applications[1]. The ends were the client and the server and the data was encrypted "end to end". The goal at that time was to prevent MITM snooping/attacks which were highly prevalent at the time.
Papers in academia and the greater industry[2] also referred to it in this way at the time.
Stack Overflow has plenty of examples of folks calling it "end to end encryption" and you can start to see the time period after the Signal protocol and WhatsApp implemented it that the term started to take on a much wider meaning[4]
This also came up a lot in the context of games that rolled out client side encryption for packets on the way to the server. Folks would run MITM applications on their computer to intercept game packets coming out of the client and back from the server. Clever mechanisms were setup for key management and key exchange[3].
[0] as SSL became more common lots of tooling broke at the network level around packet inspection, routing, caching, etc. As well as engineers "having fun" on Friday nights looking at what folks were looking at.
[1] Stack Overflow's security section has references from that era
At least in some circles, the real meaning of "end-to-end encryption" was being addressed. For example, in the field of credit card processing, here's an article from 2009 which talks about how people back then were misusing the term: https://web.archive.org/web/20090927092231/http://informatio...
Granted, it's a marketing piece trying to sell a product, but still.
I wasn't a network engineer, but to my recollection "end-to-end encryption" was only used occasionally, probably by people not too knowledgeable in cryptography
Well respectfully your recollection is missing lots of references by people that were "knowledgeable in cryptography".
You can easily find these references in the literature, often comparing link encryption with end-to-end encryption. Some of the earliest papers outlining the plans for SSL in the 90s (Analysis of the SSL 3.0 Protocol) are based on this exact foundation from the 80s (End-To-End Arguments in System Design).
Hell, you can even go back to 1978 and see MITRE discussing this exact thing in "Limitations of end-to-end encryption in secure computer networks".
With three citations I was about to give in, and accept that my experience might have been limited, but then I checked those citations and... are you trolling? Or were those given you by an llm?
1. "End-To-End Arguments in System Design" (https://web.mit.edu/Saltzer/www/publications/endtoend/endtoe...) argues that it's appropriate to perform various functions at the high-level, application, ends, rather than for example leaving encryption to devices external to the hosts.
It's really a stretch to affirm that it considers "end-to-end encryption" a proper term for transport, client-server encryption.
Actually, I'd say that transport-level, origin-server -> server-destination encryption is precisely one of the things that the paper would not consider end-to-end.
a. it doesn't "outline the plans for ssl", it's an analysis of its third version???
b. It doesn't reference "End-To-End Arguments in System Design" anywhere, and doesn't even contain the expression "end-to-end"
3. "Limitations of end-to-end encryption in secure computer networks" is mostly concerned with warning about side-channels, that they can be used to disseminate information despite encryption.
Its usage of end-to-end encryption is defined in the paper that's being criticized (https://dl.acm.org/doi/pdf/10.1145/1499799.1499812):
«The term end to-end encryption refers to data being enciphered at the source and remaining unintelligible until it deciphered at its final destination.»
I'll take the hit on the loose phrasing regarding the SSL paper "outlining plans". That was a poor description of mine of an analysis paper and wasn't a good example of the point I was trying to make. However, you are focusing on the trees and missing the forest. The citations you analyzed actually prove the semantic shift I am describing, specifically the MITRE one.
You quoted the MITRE paper (or the older paper it references) defining end-to-end encryption as:
> "data being enciphered at the source and remaining unintelligible until it deciphered at its final destination."
This is the exact crux of the disagreement. In classic Client-Server architecture, the Server was the "final destination". The application processing the data lived on the server. Therefore, by the definition you just quoted, SSL/TLS from Client to Server was "End-to-End Encryption" because the network (routers/ISPs) could not decipher it.
The "modern" definition (post-Signal/WhatsApp) effectively redefined "final destination" to mean "another human user," relegating the Service Provider to a mere hop in the middle. That is a massive semantic shift.
re Saltzer's "End-to-End Arguments": The paper argues that functions (like reliability or encryption) should be moved from the lower network layers (links) to the "ends" (hosts/applications). SSL/TLS is the literal implementation of this argument: moving encryption out of the network links (Link Encryption) and into the application endpoints (Host-to-Host).
The term "End-to-End" in networking *has* historically meant Host-to-Host (Transport Layer), whereas the modern messaging usage means User-to-User. That is why a lot of folks from that era (and the RFCs) called SSL "End-to-End encryption" because relative to the network, it is.
> At this time all Internet Protocol (IP) packets must have most of their header information, including the "from" and "to" addresses, in the clear. This is required for routers to properly handle the traffic even if a higher level protocol fully encrypts all bytes in the packet after the IP header. This renders even *end-to-end encrypted* IP packets subject to traffic analysis if the data stream can be observed.
---
Regarding your claim that "no one really used the E2EE term before it got the current meaning," the IETF standards for the internet (albeit an informational RFC and not a standards RFC) explicitly list SSL and TLS as examples of End-to-End encryption. The definition of "End" has simply shifted from the Machine to the User.
> I'll take the hit on the loose phrasing regarding the SSL paper "outlining plans". That was a poor description of mine of an analysis paper and wasn't a good example of the point I was trying to make
I don't understand why you cited it at all; I didn't read it carefully, but I didn't find anything relevant to the discussion.
---
RFC4949 might indeed support your point; it says intended final destination, though: while SSL is listed among the examples, does that include the "SSL-server-SSL" of a non-E2EE messaging system?
I think there's a good chance that it doesn't, in the intentions of the RFC's authors.
---
> This is the exact crux of the disagreement. In classic Client-Server architecture, the Server was the "final destination"
The disagreement is on whether in a user-server-user system, encrypting the two user-server sides was ever considered sufficient to call it an end-to-end encrypted system.
I think it wasn't, and to my recollection, luckily, no one ever tried to call it that.
Keep in mind that it used to be rare both to use any kind of encryption, and to go through an intermediary server for real-time, one-to-one communication.
It's only when centralized messaging systems begun to use SSL that the possibility of confusion arose.
They should just never have called themselves encrypted, in my opinion; encrypting the traffic was sure a big improvement, but I'd only call a messaging system encrypted if no decryption occurs before reaching the recipient
---
> The definition of "End" has simply shifted from the Machine to the User.
The ends are actually machines in the current definition too, it's not like people decrypt stuff by hand ;)
---
You sure proved that E2EE was a term already in use, anyhow (although I don't think too widely)
The two endpoints of the communication with Kohler's app are the client and the server. In WhatsApp's E2EE implementation the endpoints are two client devices. Both are valid meanings of E2EE. You're defining that "end to end" means the server cannot access it but that's simply not what it means.
The modern usage of E2EE definitely means that "the server cannot access it". That's the meat of this entire discussion.
While you are technically correct in a network topology sense (where the "ends" are the TCP connection points), that definition has been obsolete in consumer privacy contexts for a decade now due to "true" E2EE encryption.
If we use your definition, then Gmail, Facebook, and Amazon are all "End-to-End Encrypted" because the traffic is encrypted between my client and their server. But we don't call them E2EE because the service provider holds the keys and can see the data.
In 2025, when a company claims a camera product is "E2EE", a consumer interprets that to mean "Zero Knowledge". I.e. the provider cannot see the video feeds. If Kohler holds the keys to analyze the data, that is Encryption in Transit, not E2EE. Even though in an older sense (which is what my original comment was saying), it was "End to End Encrypted" because the two ends were defined as Client and Server and not Client to Client (e.g. FB Messenger User1 and FB Messenger User2).
> If we use your definition, then Gmail, Facebook, and Amazon are all "End-to-End Encrypted" because the traffic is encrypted between my client and their server.
That may or may not be the case. TLS is always terminated at a load balancer that uses TLS but it's still common to use HTTP within datacenters. So it may not be E2EE and it's a meaningful security feature.
No term will stop marketers from lying. If users see one as being the more secure one, marketers will use it. Unless they get sued for false advertising.
> However in this case there are no other users, and their server is one of the "ends" doing the communicating, which is... perhaps not a literal contradiction in terms, but certainly breaking the spirit of the phrase.
Am I understanding correctly that the other end of this is a rear end?
> They're claiming "end to end" encryption, which usually implies the service is unable to spy on individual users that are communicating to one-another over an individualized channel.
It doesn't "imply", it outright states that. Their server isn't the end, it's the middle. They're not "breaking the spirit" or something, what they are doing is called lying.
This is exactly what E2EE means. I used to work at a bank, and our data was E2EE, and we had to certify that it was E2EE - from the person paying, through the networks, through the DNS and Load balancers, until it got to the servers. Only at the servers could it be unencrypted and a (authoried) human could look at it.
Of course, only authorized users could see the data, but that was a different compliance line item.
No, E2EE doesn't mean it's encrypted until the service provider decrypts it. E2EE means the service provider is unable to decrypt it. What you are describing is encryption in transit (and possibly at rest).
Bank data is never E2EE because the bank needs to see it. If banks call it E2EE they are misusing the term. E2EE for financial transactions would look like e.g. ZCash.
I would argue it depends on context. E2EE means it's encrypted until the "target" receives it. For a messaging protocol, it's the intended recipient of the message. For what the person you're replying is discussing, the intended recipient IS the bank.
That being said, the person you're replying to seems to be saying that "the server" is always an "intended" end, which is wrong.
No, it doesn't depend on context. The intended recipient of a financial transaction is not the bank. The intended recipient is the party you're trying to pay. It is possible for financial transactions to be E2EE and completely indecipherable by anyone but the two parties of the transaction. Crypto like ZCash can do it. Banks cannot.
Can you expand on this a bit. It was my understanding that you're telling the bank to pay the vendor (from your money/credit). In that case, the bank certainly needs to know about the transaction... so it can make the payment.
I suggest researching how ZCash uses zero-knowledge proofs to allow paying money from your balance to another person's balance without any middleman like a bank being able to decrypt your transaction, while still allowing everyone to verify that important invariants are maintained, such as not allowing you to spend more money than you have.
This is what it takes to make a financial transaction E2EE. I'm not saying that banks could or should do this. I'm just saying that their systems do not qualify as E2EE unless they do. It's not ambiguous.
Doesn't the anonymous-ness of crypto/zcash make it impossible for the bank to handle fraud (reversing of charges and such)?
My understanding is that banks, at least in the US, need to have fairly extensive knowledge relating to all transfers of money, both for fraud handling and for non-fraud (money laundering, etc). A transaction they can't know anything about other than "transfer X money to some recipient you can't know anything about" just doesn't seem realistic with the regulations involved.
Plus, even "transfer X money to some recipient you can't know anything about" is a message that you're sending _to_ the bank, that they have to be able to decode and read. And, presumably, you'd encrypt that message and expect the bank to decrypt it.
Honestly, I don't understand what argument is that you're not sending a message TO the bank, and they need to be able to read it in order to act on it, and they need to decrypt it to read it. The bank is the target of the message, they are one of the "ends" in E2EE.
I feel like I need an "Explain this like I'm 5", because clearly you believe differently than me... and I don't understand _how_ it can be otherwise.
> Honestly, I don't understand what argument is that you're not sending a message TO the bank, and they need to be able to read it in order to act on it, and they need to decrypt it to read it. The bank is the target of the message, they are one of the "ends" in E2EE.
You might just as well say that E2EE messaging is impossible because you are sending a message "to" Signal, and they need to read it in order to act on it.
I'm telling the bank "I want you to give $5 of my money to Bob". I'm not asking them to pass a message to Bob. The entire message is the instructions for the bank to give the $5 to Bob. The bank MUST be able to read that message in order to follow the directions. There's nothing to "leave encrypted" to treat the bank as a non-end of the E2EE.
You could presumably hide who Bob is by making it some kind of anonymous account thing... but that _still_ wouldn't leave any message encrypted. Because all of the information needs to have been decrypted for the bank to act on it.
For the Signal analogy to apply, there would need to be some message going to Bob. And there isn't... other than "We're giving you this $5 for OP", all of which is information in the original message that the bank needs to act on it.
> I'm not saying that banks could or should do this. I'm just saying that their systems do not qualify as E2EE unless they do. It's not ambiguous.
That said, it might not be impossible to implement some enforcement of AML-like rules with zero-knowledge proofs. What's possible with advanced cryptography is not at all intuitive. But banks profit from their middleman position and surely wouldn't be interested in disintermediating themselves. Neither would crypto people be interested in adding AML. So I don't expect anyone to try. This fact still doesn't make existing middleman banks qualify as E2EE.
While what you're saying makes sense, it's not the normal use of the term - in fact, the term 'end to end encryption' was basically coined to differentiate user-to-user encryption (through an intermediary service that can't decrypt the message) from the regular case (user to service encryption) that you're talking about!
$ end-to-end encryption
(I) Continuous protection of data that flows between two points in
a network, effected by encrypting data when it leaves its source,
keeping it encrypted while it passes through any intermediate
computers (such as routers), and decrypting it only when it
arrives at the intended final destination. (See: wiretapping.
Compare: link encryption.)
Examples: A few are BLACKER, CANEWARE, IPLI, IPsec, PLI, SDNS,
SILS, SSH, SSL, TLS.
Tutorial: When two points are separated by multiple communication
links that are connected by one or more intermediate relays, end-
to-end encryption enables the source and destination systems to
protect their communications without depending on the intermediate
systems to provide the protection.
There's a bunch of older references as well. Since SSL/TLS wasn't really adopted by a lot of services until 2008+ usages of it are mainly in papers, old forum posts, etc. I saw it used and was discussing it back in the day on IRC with folks who were way more knowledgeable than me on this topic and had been in the trenches for a while :D
Nah. You have no reasonable expectation that the bank itself can’t access your financial records. Anyone reading Kohler’s lies would have every expectation that the Internet of Poopcam screenshots are theirs and theirs alone.
Anyone reading that is misunderstanding what E2EE means. As the article says, that's client-side encryption. Kohler isn't lying, people are confusing two different security features.
They're also claiming regulatory requirements as features. At least consumers might be able to sue in addition to several governments when it turns out to be a bunch of crap.
They're claiming "end to end" encryption, which usually implies the service is unable to spy on individual users that are communicating to one-another over an individualized channel.
However in this case there are no other users, and their server is one of the "ends" doing the communicating, which is... perhaps not a literal contradiction in terms, but certainly breaking the spirit of the phrase.