Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > sci.crypt > #32612

Re: What can an attacker see in an encrypted message?

Show key headers only | View raw

[At the risk of repeating what others have said, let me repeat
what others have said]

On 2015-10-05 12:02:46 +0000, Anonymous Remailer (austria) said:

> After seeing all of the cryptanalysis on the ng recently it occurs
> to me to ask, [...]

You are seeing cryptanalysis of "pre-modern" ciphers. The kinds of
things in the "Happy Birthday" puzzle uses design principles from 100
years ago only (other than the fact that it operates on bytes (or
perhaps bits) instead of on letters.

All of the flaws in the toy ciphers that are being exploited are
things that the world moved away from once it became abundently clear
how easily they can be broken.

> just how much can an attacker figure out from a ciphertext? What if
> he or she has some of the passphrase? Here I'm talking about a
> ciphertext created using one of the algorithms available in gpg.

A key (or a passphrase) is an "all or nothing thing" in a modern
cipher (like those used by GPG). If the passphrase is off by a single
bit, decryption fails completely.

Of course an attacker who knows most of your password might take
guesses at the remaining portions, but there is no sign or
information leak from getting close to the right passprhase.

> And, pretty much on this same subject, could an attacker tell what
> kind of data is encrypted in a file? Can they tell if it's a binary
> or a plaintext file that's encrypted?

They can tell its (approximate) length. And they may be able to make
some guesses based on things other than the actual content. For
example, an encrypted file named

 my-plan-to-shoot-the-president.docx.gpg

might provide some hint that it is a Microsoft Word document.

> Can they tell much about the contents of the ciphertext at all
> without having part of the passphrase, or a part of the session
> key?

Again, no. Getting the right key is an "all or nothing" thing.
However, let me state again that if, say, an attacker has 88 bits of
a 128 bit key, then they only have to guess at the remaining 40 bits,
testing each one. Such an attack is definitely feasible.

So you don't want part of your key or part of your passphrase to be
known. But there is no way for an attacker to know that they have
"60% of the key". A key that is a little bit wrong produce totally
wrong results.

> Could the attacker figure out a frequency of characters, or tell
> that some particular characters are used in a ciphertext without
> having the passphrase or session key?

No. Not for any serious cipher deployed in the past 40 years.

> Would a nation state attacker be able to figure out much from a
> ciphertext created using one of the algorithms available in gpg?

<PedanticMode>
"Nation state" is not the appropriate term for "country." It is a
term that applied to those countries that were created as being
states for a particularly "nationality", and so does not apply to any
multi-ethnic country.
</PedanticMode>

The short answer to your question is "No".

There is no reason to believe that any government, including the US,
is capable of performing a partial break on something encrypted with
PGP/GnuPG.

But

... A motivated attacker willing and able to put lots of resources
into learning the content of such traffic may well be able to learn
stuff by going around the crypto. Generally the cryptography is the
strongest part of the system, so an attacker will go around it. This
may be by tampering with the computers of some of the parties; going
after the message either before it is encrypted or after it is
decrypted; installing monitoring equiptment near the targets (for
example a camera that watches a computer screen), etc.

Quite simply, an attacker would rather spend tens of thousands of
dollars going around the encryption than tens of millions in a vain
attempt breaking the crypto.


And (as others pointed out), there is "traffic analysis"

Even though the contents of a message are not leaked, the existence
of a message is.

But to return to your original question, when something is properly
encrypted with a cipher designed in the past 40 years, there is no
way to draw any conclusion (other than its existence and length)
about the content of the message from its ciphertext,

Back to sci.crypt | Previous | Next — Previous in thread | Next in thread | Find similar | Unroll thread

Thread

What can an attacker see in an encrypted message? "Anonymous Remailer (austria)" <mixmaster@remailer.privacy.at> - 2015-10-05 14:02 +0200
  Re: What can an attacker see in an encrypted message? Richard Heathfield <rjh@cpax.org.uk> - 2015-10-05 13:39 +0100
  Re: What can an attacker see in an encrypted message? Paulo Marques <pmarques@grupopie.com> - 2015-10-05 14:19 +0100
  Re: What can an attacker see in an encrypted message? William Unruh <unruh@invalid.ca> - 2015-10-05 20:46 +0000
  Re: What can an attacker see in an encrypted message? Robert Wessel <robertwessel2@yahoo.com> - 2015-10-05 22:39 -0500
    Re: What can an attacker see in an encrypted message? Jakob Bohm <jb-usenet@wisemo.com> - 2015-10-06 08:01 +0200
      Re: What can an attacker see in an encrypted message? PA7PTL <PA7PTL@panteltje.com> - 2015-10-06 07:07 +0000
        Re: What can an attacker see in an encrypted message? Ben Bacarisse <ben.usenet@bsb.me.uk> - 2015-10-06 11:22 +0100
          Re: What can an attacker see in an encrypted message? eugene.styer@gmail.com - 2015-10-06 08:48 -0700
    Re: What can an attacker see in an encrypted message? Richard Heathfield <rjh@cpax.org.uk> - 2015-10-06 10:15 +0100
  Re: What can an attacker see in an encrypted message? Jeffrey Goldberg <nobody@goldmark.org> - 2015-10-06 11:22 -0500
    Re: What can an attacker see in an encrypted message? Jakob Bohm <jb-usenet@wisemo.com> - 2015-10-06 20:12 +0200
      Re: What can an attacker see in an encrypted message? Jeffrey Goldberg <nobody@goldmark.org> - 2015-10-07 19:21 -0500
        Re: What can an attacker see in an encrypted message? Richard Heathfield <rjh@cpax.org.uk> - 2015-10-08 06:33 +0100
          Re: What can an attacker see in an encrypted message? Jeffrey Goldberg <nobody@goldmark.org> - 2015-10-08 22:21 -0500
            Re: What can an attacker see in an encrypted message? Richard Heathfield <rjh@cpax.org.uk> - 2015-10-09 07:34 +0100
  Re: What can an attacker see in an encrypted message? Anonymous <anonymous@foto.nl1.torservers.net> - 2015-10-07 09:20 -0400

csiph-web