From: Rich Graves <llurch@networking.stanford.edu>
Date: Tue, 9 Jan 1996 11:27:37 +0800
To: cypherpunks@toad.com
Subject: Microsoft continues to mislead public about Windows security bugs (a bit long, with references)
Message-ID: <Pine.ULT.3.91.960108181532.24578C-100000@Networking.Stanford.EDU>
MIME-Version: 1.0
Content-Type: text/plain


-----BEGIN PGP SIGNED MESSAGE-----

Please do not dismiss this as mere "Microsoft Bashing." c2.org has
similar promotions running for Netscape, DigiCash, and Java. 

The following is a quote from Microsoft's "Knowledge Base" technical
support and marketing database, which is online in CompuServe and at: 

  http://www.microsoft.com/kb/peropsys/windows/q90271.htm

  Security of the Windows for Workgroups Password Cache
  _____________________________________________________

  The password list file is encrypted with an algorithm that meets the U.S.
  government Data Encryption Standard (DES). This encryption technology is
  the highest security allowed in software exported from the United States.
  The odds of breaking the encryption algorithm are less than those for
  random guesses of what the password might be.

  Even if your logon password is blank, Windows for Workgroups generates
  seemingly random data in your PWL file, so you cannot discover the
  passwords if you look at the PWL file using a file viewer. Currently, no
  user interface exists that allows you to unencrypt passwords in the PWL
  file, so password caching in Windows for Workgroups is as secure as the
  choice of the password used to encrypt your PWL file.

As Microsoft well knows, this is completely untrue. The rest of the world
has known that this is untrue since November 29th. Microsoft quietly
acknowledged on December 7th (after a day of much "Internet Strategy"
hype, and after the deadline for the morning papers) that the exact same
implementation was insecure in Windows 95, and claims to have released a
patch that fixes the problem (the efficacy of the Win95 patch does not
appear to have been verified by anyone outside Microsoft, however). 

Microsoft has not even admitted that this bug in both Windows 95 and
Windows for Workgroups affects Windows for Workgroups, apparently because
they have decided not to fix it. 

Information on the .PWL implementation bugs was first broached on the 
sci.crypt newsgroup in late November 1995, then discussed on the 
cypherpunks list and refined for Community ConneXion's "Hack Microsoft"
promotion, http://www.c2.org/hackmsoft/.

We have since been given a sample trojan horse that will very efficiently
exploit this bug in Windows for Workgroups. Distributed as a Word Basic
virus, MIME attachment, or downloadable archive (note that Exchange and
Internet Explorer unwisely execute downloaded binaries without even a
virus check, a problem that Sun's Java has long acknowledged and
addressed), this trojan horse could collect passwords and other sensitive
information from .PWL files and other sources and send them out via email,
possibly through an untraceable chain of remailers or to a throwaway
trial account on, for example, America "Online." 

We believe that it would be highly irresponsible to release the full
version of this hack, but we will soon release a crippled
demonstration-only version if Microsoft does not at the very least admit
that this problem has always affected Windows for Workgroups, correct
their online documentation, publish the specifications of the Win95
security patch for review by outside security experts, and issue a public
retraction. 

See also:

  http://www.microsoft.com/kb/peropsys/windows/90210.htm
  http://www.microsoft.com/windows/pr/clarifications.htm
  http://www.c2.org/hackmsoft/

  http://www-leland.stanford.edu/~llurch/win95netbugs/faq.html

  {mirror of above} http://www.mari.su/guide/win95/
  {mirror of above} ftp://ftp.demon.co.uk/pub/mirrors/win95net/
  {more mirrors are under construction in Australia and elsewhere}

In other news, I assume everyone knows by now that NT's claimed C2
security rating was granted *for use a standalone workstation only*. It
has been widely reported that its NetWare Services implementation does
not ask for passwords for nonexistent usernames, making a potential
cracker's job that much easier. The correct response, which is given by
real NetWare servers and other servers that are certified C2-secure on
networks, is to silently ask for a password in all cases. 

I started getting copies of hackmsoft@c2.org mail on December 20th. It's 
really depressing.

We've also seen problems with Microsoft Access 95's security. Basically,
there is none. Anyone can access the network-enabled Access as any user
without knowing the password. We don't think it would be responsible to
publicly release this hack, either, until Microsoft has had another
chance to patch the hole (they've known about it for some time). 

These are far, far worse than the widely publicized bugs in Netscape's
SSL implementation, which have been fixed. Yet the only place I've seen
them mentioned is the lapdog Seattle Times, which only reports bug
*fixes* in glowing terms. 

Is anybody listening?

- -rich
 owner-win95netbugs@lists.stanford.edu
 ftp://ftp.stanford.edu/pub/mailing-lists/win95netbugs/
 gopher://quixote.stanford.edu/1m/win95netbugs
 http://www-leland.stanford.edu/~llurch/win95netbugs/faq.html

-----BEGIN PGP SIGNATURE-----
Version: 2.6.2

iQCVAwUBMPHZrI3DXUbM57SdAQFojwP/T0CIjfyEz5NHD81wPdkAuUf1YCB8OE3/
4NakffTxzmPxJXRT/MoRpOMn4qJa6mzC6WAgAdwtKWG/3K9WS1LNgM/w/PYMHj45
pEQroJBzoXU/Sctjnyz87FBl2/m6dwAdvPQqGOzGqsLVDaFsmqbWtalkvP2y0707
ntdb2fkqpNI=
=q491
-----END PGP SIGNATURE-----