gpg4win

[none]

Is this application regarded as being secure.
Comments requested.
Thank you.

[none]

[Kristian Gjųsteen]

none <no...@such.com> wrote:
>Is this application regarded as being secure.

It is linked to from the front page of www.gnupg.org. It seems to be
a front-end for GnuPG. It should be secure, since most of the really
difficult stuff is done by GnuPG, which seems to be well regarded.

--
Kristian Gjųsteen

[Harris]

Kristian Gj?steen <kristi...@math.ntnu.no> wrote in news:h0r9v4-pqe.ln1@fimf-
h28.math.ntnu.no:

I've been using gpg4win and works fine for me. On the securoty part, it's really gpg code underneath,
just compiled for win32. I haven't analyzed the sources myself, but it's considered as secure as the
original gpg for Unix/Linux platforms (e.g. no evident backdoors). It's also compatible with most pgp
versions, so keyring import/export is quite easy between the two.

The only problem is that the provided plugin for Outlook (which is the most common mail client for
win32) seems to crash 99% of the times, in other words no embedded email support (has to be donw
via clipboard or files).

--
Harris

[frankg...@gmail.com]

The security of the computer is probably more critical than that of
any encryption program.
If one is really serious about confidentiality of data &
communication, probably two
computers should be used: one which is connected to the internet (and
possibly
compromised by all kinds of hackers or agencies) and another computer
that
does the en/decryption (using a strong tool such as gpg) and stores
all confidential
data. This method is also called an "air gate". Note that it is not
totally safe, as
complex document formats such as Office, PDF of html could contain
malicious
code. There are just too many buffer overflow exploits in popular
commercial
(and open-source) software. If you habe something really confidential,
resort
to ASCII....

[frankg...@gmail.com]

The "air gate" term refers to the fact that the "online" and the
"crypto" computer are not
connected via an automated mechanism (such as ethernet or WLAN). Data
is only "transmitted" via
storage media such as DVDs, floppys or a USB stick.
Which might open an attack venue in the form of a buffer overflow
attack into the file system code :-(

[Simon Johnson]

frankg...@gmail.com wrote:

I've given a similar set-up considerable thought.

You could make a very stripped down distribution of Linux that contains only
a basic shell, basic file manipulation utilities and the encryption
program.

This distribution would boot off CD on a machine with no hard-drive. The
distribution would have no networking capability compiled in to the kernel.

The one point where I disagree with what you've proposed is that the
plain-text be authored on the hostile machine. I think this is a bad idea
because that very machine is considered hostile.

If the plain-text is what they're after then they'd just target the Internet
facing machine which we've defined to be impossible to properly secure!

I think you have to compose the message on your secure machine. You compose
your message on the machine using something like Vim and save it to the ram
drive. You then perform the encryption and write the encrypted message to
some sort of removable media.

You'd want this removable media to be a blank CD or any other blank media.
You don't want to put something in to that machine that already has data on
it because that opens you up to attack.

When your sending a message, you should always use new write once media as a
matter of policy.

When receiving, you have a problem of how to ensure there isn't any
malicious data trying to infect your secure machine. A virus scanner will
only detect attacks that are not specifically aimed at you. An attacker who
is specifically targeting you can always write a virus that will evade a
virus scan.

However, since the machine is not connected to anything else, it is
difficult to see how a virus could compromise the plain-text or the
underlying cryptographic keys.

Provided you always hard-reboot the machine between sending encrypted
messages and receiving them, there is limited scope for damage. In fact, as
a matter of policy I would perform a hard-reboot between the handling of
each message.

Under this set-up, I would be reasonably confident that the plain-text has
been well protected. What's nice about this scheme is that it is reasonably
cost effective. An old 200MHz PC could easily be re-purposed for this sort
of set-up. The cost of a single CD per message is reasonable if you really
value your security.

Simon

[John E. Hadstate]

"Simon Johnson" <Simon....@gmail.com> wrote in message
news:47248696$0$13923$fa0f...@news.zen.co.uk...

>
> You could make a very stripped down distribution of Linux
> that contains only
> a basic shell, basic file manipulation utilities and the
> encryption
> program.
>
> This distribution would boot off CD on a machine with no
> hard-drive. The
> distribution would have no networking capability compiled
> in to the kernel.
>

Take a look at Knoppix. To do what you suggest, you will
need their SDK. Otherwise, it's a class act.

[Dav170627]

A windows alternative could be to run PC simulator software for your
Internet connections(Microsoft make one - Microsoft Virtual PC - and
there is at least one other, last I heard they were both free)

If things go bad you just delete the virtual PC file and begin again.