back to article Can WhatsApp moderators really read your encrypted texts? Yes ... if you forward them to the abuse dept

Facebook's WhatsApp states its messages are protected by the Signal encryption protocol. A report published today by investigative non-profit ProPublica contends that WhatsApp communication is less private than users understand or expect. "WhatsApp assures users that no one can see their messages — but the company has an …

  1. Il'Geller

    AI technology and absolute censorship, no escape from this

    AI technology assumes absolute censorship, and there will be no escape from this. The AI understands all the meaning, not the words: enough to create a specifically oriented profile, as it will automatically catch 100% of certain information. This means total control over what is happening on the Internet. The Internet has become a database where everything is laid out on the right shelves, numbered and sorted by meaning. The new era, the new rules!

    1. elsergiovolador Silver badge

      Re: AI technology and absolute censorship, no escape from this

      AI can't reason or think. It is just a sophisticated pattern matching. It cannot tell whether a message is sarcastic or serious or a joke. Frankly, not many people can do that either...

      That explosion of AI technology is to an extent a scam. Many people can be fooled by the results and easily part with their money thinking they struck a gold mine. Only to have their happy lives traded to constantly checking news how that training of models is going and how data "scientists" need more data or need to do more "tweaks" they cannot even explain :-)

      The worst thing is that lazy government bodies are buying into that idea that AI could do their jobs for them while they can just work from home and eat cookies whole day. To an extent it's true, AI could do the job of many of them, but so could my cat.

      1. Martin Summers Silver badge

        Re: AI technology and absolute censorship, no escape from this

        "Many people can be fooled by the results"

        And indeed I think you've just proven that.

    2. Throatwarbler Mangrove Silver badge
      Facepalm

      Re: AI technology and absolute censorship, no escape from this

      Don't you have a winch to wind?

  2. elsergiovolador Silver badge

    An invisible hand of censor

    So if there is a mechanism that allows sending an unencrypted message to "moderators", can now government ask Whatsapp to send any message on user's behalf?

    1. Androgynous Cupboard Silver badge

      Re: An invisible hand of censor

      No.

  3. Throatwarbler Mangrove Silver badge
    Facepalm

    Disappointing indeed

    Normally ProPublica does much better work. I can't believe they couldn't figure out the fact that, if you forward an encrypted message to someone, then the new recipient can read the forwarded message.

    1. Dinanziame Silver badge

      Re: Disappointing indeed

      It might be technically a bit more complicated. If I just forward a message, it would be trivial for me to modify the message. Or create a fake snapshot, or something similar. Here, I assume WhatsApp is at least able to confirm the original message. It's also possible — maybe, I don't know — that rather than forwarding the message, the reporting user transmits the decryption key to allow the abuse team to read the original message directly.

      Ultimately, you can be spied on, even in the real world. The UK is a good example.

    2. Anonymous Coward
      Anonymous Coward

      Re: Disappointing indeed

      Particularly for abuse reports. If they can't read the message that was reported as abuse, how can they tell if it was abuse? Turn off that capability, and there's no longer a way to kick bad actors off the system, since they can't be verified as bad actors.

      In other words, this is intentional and expected reading of messages by WhatsApp, at the specific request of the recipient. But only the messages that they asked WhatsApp to review.

  4. Anonymous Coward
    Anonymous Coward

    I always wondered about how end-to-end the encryption really is.

    These messaging apps uses a text box from whatever interface toolkit the platform uses, which you type in plain text and I imagine in you'd do inputbox.value on pressing the submit button to get the text you've entered and then pass it as a parameter to libsignal to handle the encryption and sending, so what would stop whatsapp assigning other events to the submit button and sending those messages elsewhere or is there some "magic" which prevents the developer from getting the plain text from the message input box? After looking at libsignal you do in fact pass it some plain text.

    1. Richard 12 Silver badge

      Well, obviously?

      For a start it's a touch keyboard, so the keyboard itself could be compromised.

      There have been well publicised incidents where the endpoint was compromised to monitor messages sent and received by that endpoint passing them on elsewhere.

      One thing that is not clear is what is actually sent when you make an abuse report. Are you giving them the last five messages, last five days, back to the dawn of time, or even access to future messages?

      Eg if it sends the key then your report effectively contains everything, both past and future.

      1. Anonymous Coward
        Anonymous Coward

        No my point is whats stopping whatsapp simply sending a second request using their key which of course they can decrypt as both messages will be going through the same server, who would know? Facebook at that point can read your conversation.

    2. Anonymous Coward Silver badge
      Holmes

      The "end" in end-to-end-encryption is somewhere in the software stack of your device.

      Anything between that "end" and the user's eyeballs (and I'm not convinced it stops there) could be compromised to reveal the contents of the messages. That's by design. To use encryption that survives all the way to the user's brain, you'd have to learn Welsh.

      1. EricB123 Bronze badge

        Welsh?

        Is Welsh THAT hard to learn?

  5. Kevin McMurtrie Silver badge

    Prodigy, AOL, Angelfire, Geocities, Yahoo, Google

    The obvious solution is to stop taking abuse reports. It's expensive and the exec staff can retire before the imminent failure.

  6. Anonymous Coward
    Meh

    A bit over the top?

    It's a bit of an odd attack - I'm usually in favour of trashing Facebook, but it seems that conversations are private, subject to the proviso that you can't control what someone does with your message at the other end. Which applies equally to Signal, for example.

    Where Facebook fall down, maybe, is that in stating that conversations are private they are brushing over that point, leaving the unwary to assume absolute privacy. I'm not sure that Facebook are exactly alone in doing this.

    1. big_D Silver badge

      Re: A bit over the top?

      The messages are private, unless the recipient is offended by what you send them and pass the message on to the abuse reporting team...

      Or they make a snapshot and post it online, or they copy and paste it into a blog post or into another WhatsApp conversation...

      At the end of the day, the messages are encrypted at rest, encrypted in transit, but open for the user to see and the user can do anything they like with the message, once it has been displayed.

  7. bluesxman

    Seems pretty obvious

    As otherwise commented, it seems to stand to reason that a message forwarded to the Abuse account would be visible to that account, much like a forwarded email is obviously visible to the recipient. They couldn’t realistically act on such a report with seeing it.

    I’m tacitly assuming that the integrity of such a message is protected by attached meta data. Like, some sort of hash is produced from the original message (and/or it’s component parts) at the point of sending, prior to encryption. Such hash(es) would then be encapsulated with the encrypted message. Thus the message can be uniquely identified but actual content not known unless the Abuse account receives it along with the meta data. The hash could then be trivially used to confirm that Bob received the message from Alice and it was passed in unaltered. This hash could also be used to identify/filter other identical instants of the offending message sent by Alice or any given user.

    If Bob changed the message before reporting, the hash of his report would differ and the chain of custody would be broken.

    There are just my musings about how that sort of thing could work, I have no knowledge of the inner workings.

    1. bluesxman

      Re: Seems pretty obvious

      Sorry a few typos there, I blame the proof reader. Hopefully you can derive intended meaning.

    2. big_D Silver badge

      Re: Seems pretty obvious

      Yes, I think Facebook and its properties are the spawn of Satan and deserve to burn in a pit of hell...

      That said, I stick up for them in this case. The ProPublica report is just click-bait and is a misrepresentation of the facts.

      How else is an abuse investigation team going to investigate, if they can't see the abusive messages that have been reported to them?

      And, unless they are lying and don't use the Signal protocol, they can't see every message, they can only see the messages that an end user has explicitly sent to them as an abuse report.

    3. Pascal Monett Silver badge

      Re: Seems pretty obvious

      As far as I can tell, a forwarded message is a new message.

      Entirely new, with content copied from a given message, of course, but there is otherwise no link to the original message that has been forwarded.

      So receiving an encrypted message that only you can read is one thing. Forwarding that message to someone else creates a new message with a different encryption value to the new recipient.

      If the writer of the original message somehow gets a copy of the forwarded message, he should not be able to read it because he is not the recipient.

      Unless, obviously, the recipient of the forwarded message forwards it to him.

      Am I being clear ?

  8. DevOpsTimothyC

    Poor El-Reg reporting

    Adding the proton mail element isn't really relevant to WhatsApp. In WhatApp you have users reporting abuse in proton mail there is a court order.

    Of course in the proton case their (French) T&C's denied that the court order could get the information that was handed over. The user was French.

    1. lglethal Silver badge
      Go

      Re: Poor El-Reg reporting

      It's not bad reporting at all. In this article, the Proton case was brought up to discuss the point about the perceived difference between the T&Cs from the consumer point of view and from the company's point of view.

      It seemed quite a good example of potential consumer misunderstanding, or the lack of clarity in privacy claims amongst tech companies.

  9. mark l 2 Silver badge

    I use Whatsapp quite frequently and wasn't even aware there was a reporting option, After reading this article I decided to look for it and its hidden away in the more options from the menu, so I doubt it gets used very often.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like