![[LWN Logo]](/images/lcorner.png) |
|
![[LWN.net]](/images/Included.png) |
From: Bruce Schneier <schneier@counterpane.com>
To: crypto-gram@chaparraltree.com
Subject: CRYPTO-GRAM, September 15, 2001
Date: Sat, 15 Sep 2001 11:41:36 -0500
CRYPTO-GRAM
September 15, 2001
by Bruce Schneier
Founder and CTO
Counterpane Internet Security, Inc.
schneier@counterpane.com
<http://www.counterpane.com>
A free monthly newsletter providing summaries, analyses, insights, and
commentaries on computer security and cryptography.
Back issues are available at
<http://www.counterpane.com/crypto-gram.html>. To subscribe or
unsubscribe, see below.
Copyright (c) 2001 by Counterpane Internet Security, Inc.
** *** ***** ******* *********** *************
In this issue:
11 September 2001
NSA's Dual Counter Mode
Corrections and Explanations
Crypto-Gram Reprints
News
Counterpane Internet Security News
The Doghouse: Shyfile
New Microsoft Root Certificate Program
Comments from Readers
** *** ***** ******* *********** *************
11 September 2001
Both sides of the calendar debate were wrong; the new century began on 11
September 2001.
All day I fielded phone calls from reporters looking for the "computer
security angle" to the story. I couldn't find one, although I expect
several to come out of the aftermath.
Calls for increased security began immediately. Unfortunately, the
quickest and easy way to satisfy those demands is by decreasing
liberties. This is always short sighted; real security solutions exist
that preserve the free society that we all hold dear, but they're harder to
find and require reasoned debate. Strong police forces without
Constitutional limitations might appeal to those wanting immediate safety,
but the reality is the opposite. Laws that limit police power can increase
security, by enforcing honesty, integrity, and fairness. It is our very
liberties that make our society as safe as it is.
In times of crisis it's easy to disregard these liberties or, worse, to
actively attack them and stigmatize those who support them. We've already
seen government proposals for increased wiretapping capabilities and
renewed rhetoric about encryption limitations. I fully expect more
automatic surveillance of ordinary citizens, limits on information flow and
digital-security technologies, and general xenophobia. I do not expect
much debate about their actual effectiveness, or their effects on freedom
and liberty. It's easier just to react. In 1996, TWA Flight 800 exploded
and crashed in the Atlantic. Originally people thought it was a missile
attack. The FBI demanded, and Congress passed, a law giving law
enforcement greater abilities to expel aliens from the country. Eventually
we learned the crash was caused by a mechanical malfunction, but the law
still stands.
We live in a world where nation states are not the only institutions which
wield power. International bodies, corporations, non-governmental
organizations, pan-national ethnicities, and disparate political groups all
have the ability to affect the world in an unprecedented manner. As we
adjust to this new reality, it is important that we don't become the very
forces we abhor. I consider the terrorist attacks on September 11th to be
an attack against America's ideals. If our freedoms erode because of those
attacks, then the terrorists have won.
The ideals we uphold during a crisis define who we are. Freedom and
liberty have a price, and that price is constant vigilance so it not be
taken from us in the name of security. Ben Franklin said something that
was often repeated during the American Revolutionary War: "They that can
give up essential liberty to obtain a little temporary safety deserve
neither liberty nor safety." It is no less true today.
Senate Amendment 1562, adopted Thursday, will expand Federal wiretapping
powers:
<http://www.cdt.org/security/091101response.shtml/>
Calls to ban encryption:
<http://www.wired.com/news/politics/0,1283,46816,00.html>
<http://www.msnbc.com/news/627390.asp>
Re-emergence of Carnivore:
<http://www.wired.com/news/politics/0,1283,46747,00.html>
<http://latimes.com/business/la-000073542sep12.story>
Erosions of civil liberties are coming:
<http://www.wired.com/news/politics/0,1283,46784,00.html>
Other essays:
<http://www.crypto.com/wtc.html>
<http://groups.yahoo.com/group/wtcattack/message/93>
<http://www.cdt.org/security/cdtstatement.shtml>
"Americans must rethink how to safeguard the country without bartering away
the rights and privileges of the free society that we are defending. The
temptation will be great in the days ahead to write draconian new laws that
give law enforcement agencies - or even military forces - a right to
undermine the civil liberties that shape the character of the United
States. President Bush and Congress must carefully balance the need for
heightened security with the need to protect the constitutional rights of
Americans."
- The New York Times, 12 Sep 01
"Our values, our resolve, our commitment, our sense of community will serve
us well. I am confident that, as a nation, we will seek and serve justice.
Our Nation, my neighbors and friends in Vermont demand no less, but we must
not let the terrorists win. If we abandon our democracy to battle them,
they win. If we forget our role as the world's leader to defeat them, they
win. And we will win. We will maintain our democracy, and with justice, we
will use our strength."
- Sen. Patrick Leahy, 12 Sep 01
"History teaches that grave threats to liberty often come in times of
urgency, when constitutional rights seem too extravagant to endure."
- Justice Thurgood Marshall, 1989
** *** ***** ******* *********** *************
NSA's Dual Counter Mode
Last month I mentioned that NIST is soliciting new modes of operation for
AES. One of the modes submitted was "Dual Counter Mode," by Mike Boyle and
Chris Salter of the National Security Agency. Within days of publication,
there were at least two successful cryptanalyses of the mode. (One of them
is linked to below. The other, by Phillip Rogaway, has not been
published.) Just over a week later, NSA withdrew the mode from consideration.
As cryptographers, we tend to be sloppy with the word "break." There's a
world of difference between an academic break and a practical break. I
wrote about this extensively during the AES process. Academic breaks are
the ones that force you to change things in the design process; practical
breaks force you to change things in fielded equipment. This work is
clearly of the academic-break variety.
This is not to minimize it; it's great work. Donescu, Gligor, and Wagner
broke both the security and integrity properties of the mode. That is,
they showed that the NSA claims for this mode were not true.
NSA documentation stated they have been working on this mode for 18
months. Honestly, I don't know whether to believe them. DCM is poorly
specified, a sloppiness that I would expect the NSA to have caught as the
mode moved from office to office for review.
The meta-moral here is that security proofs are important. For 20 years,
people tried to build "two-for-the-price-of-one" modes that gave integrity
for free, and essentially all such schemes were broken. As a result, the
conventional wisdom was that this was hopeless. Then Charanjit Jutla
published a paper showing that it was possible; he gave a proof of
correctness. Since then, all new "two-for-one" modes from the research
community come with proofs, whereas the NSA's mode came with no
proof. When taken in the context of all the previous failed attempts to
build such a mode, it should not be terribly surprising to find that yet
another unproven mode got broken. The lesson is clear. If you're going to
build a new mode of operation, for heaven's sake, make sure you've got a
proof of security for it.
In summary, the attacks are real and impressive, but not practical in the
real world. Still, the NSA really should have caught them. They were
right to withdraw the mode.
NIST Proposed Modes:
<http://csrc.nist.gov/encryption/modes/proposedmodes/>
NSA's Dual Counter Mode:
<http://csrc.nist.gov/encryption/modes/proposedmodes/dctr/dctr-spec.pdf>
The cryptanalysis paper:
<http://www.cs.berkeley.edu/~daw/papers/dcm-prelim.ps>
Other documentation, including NSA's retraction:
<http://cryptome.org/nsa-dcm.htm>
** *** ***** ******* *********** *************
Corrections and Explanations
I've received some reports that Dmitry Sklyarov handed out copies of
Elcomsoft's eBook-security-cracking software at DefCon. If this is true,
it certainly muddies the whole issue. The DMCA isn't any better a law, but
distributing such software within the U.S. is clearly in violation of it.
There has been a vicious rumor floating around that Ed Felten signed an NDA
before he evaluated the various RIAA technologies, and he violated that NDA
by publishing. This is completely false. The RIAA had a click-through
license on their Web page, stating that any researchers breaking the
technology would have to sign an NDA to claim the prize. Felten never
claimed the prize, and never signed an NDA. Here is the agreement:
<http://www.cs.princeton.edu/sip/sdmi/clickthru.pdf>
I've also received considerable mail about my news item poking fun at
Microsoft for sending out signed security bulletins that fail
verification. I know that PGP uses a web of trust, and that the only way
to verify a signature is to have the public key signed by some mutually
trusted party. My complaint with Microsoft is that they sign their
bulletins with a public key that is unsigned, by anybody. Hence it is
impossible to verify the signature.
One way to think of PGP is that every user is also a CA. By asking someone
to sign your key, you are asking that person to be a CA on your behalf. By
choosing to trust keys signed by someone, you are choosing to trust him or
her as a CA. When Microsoft uses a public key without any signatures, they
bypass this mechanism. You can verify the signature on Microsoft security
bulletins; the signature might be valid, but you still have no way of
knowing that the signing key actually belongs to Microsoft. You can use
some non-cryptographic means to verify they key -- you got it off the
Microsoft Web site, all Microsoft bulletins are signed with the same key,
etc. -- but that has a set of non-cryptographic risks that you may not want
to accept. Were you fooled by those fake Microsoft service bulletins or
CNN news reports that have gone by in the last year or so?
The way for Microsoft to solve this is to get their key signed by a variety
of trusted organizations, thereby getting itself immersed into PGP's web of
trust. The fact that Microsoft doesn't understand this is what I was
poking fun at.
On the other hand, someone else pointed out that SANS and CERT advisories
suffer from the same problem. Oops.
The "Code Red III" report out of Korea seems to have been a false alarm, at
least at the time I wrote that piece on Code Red last month. But since
then, there's been at least one other Code Red variant, so my point stands.
And about a zillion people sent e-mail to tell me that the Code Red worm
was not based on an eEye vulnerability announcement, but on an older worm
without a sexy name. (Oddly enough, eEye never contacted me to
complain.) I was wrong, and I unjustly laid some of the blame for Code Red
onto eEye. Apologies to them.
A much larger issue -- and also one I received a boatload of mail about --
is whether people and companies who disclose vulnerabilities are
blameworthy when hackers write exploits that target that
vulnerability. More than one person pointed out the irony of me blaming
eEye in one part of Crypto-Gram, and then praising Sklyarov in
another. One person's reaction was included as comments in a piece of
exploit code (+10 points for style, but I prefer e-mail).
In general, I am a strong proponent of full disclosure. It is a
long-standing tradition in cryptography, and has done more to improve
computer and network security than anything previously tried. However,
there is a fine line between full disclosure and arming criminals, and it's
not at all obvious where that line is at all times. Sometimes I'm not even
sure that I know. And when it is not obvious, I tend to side with openness.
I plan on talking about this in a future issue of Crypto-Gram.
** *** ***** ******* *********** *************
Crypto-Gram Reprints
Full disclosure and the Window of Exposure:
<http://www.counterpane.com/crypto-gram-0009.html#1>
Open-source and security:
<http://www.counterpane.com/crypto-gram-9909.html#OpenSourceandSecurity>
Factoring a 512-bit number:
<http://www.counterpane.com/crypto-gram-9909.html#Factoringa512-bitNumber>
** *** ***** ******* *********** *************
News
GCHQ (the British military crypto research facility that independently
invented public-key cryptography in the 1970s) has a new identity-based
public-key scheme. The details, and source code, are published.
<http://www.cesg.gov.uk/id-pkc/>
Microsoft's initial Code Red patch didn't necessarily work. I believe this
has been fixed in later versions of the patch.
<http://archives.neohapsis.com/archives/incidents/2001-08/0218.html>
This story about computerized car-security devices doesn't talk about the
safety issues. What if there's a short-circuit in the auto when the real
owner is driving, and the door-locks go into "prisoner" mode. What if the
engine-kill feature in a stolen car goes into effect when the car is on a
busy highway, simply because there's a police station nearby. I don't
think computer security is reliable enough for this kind of stuff.
<http://www.eetimes.com/story/technology/OEG20010612S0059>
Viruses and instant messaging:
<http://news.cnet.com/news/0-1005-200-6873022.html?tag=tp_pr>
A Senate report on the NSA says they're not doing well in the face of
technological advances.
<http://www.theregister.co.uk/content/55/21047.html>
<http://news.bbc.co.uk/hi/english/sci/tech/newsid_1491000/1491102.stm>
Niels Ferguson says he's broken Intel's HDCP Digital Video Encryption
System, but won't reveal details because he's afraid of being arrested
under the DMCA. I know Niels, and have always found his honesty
unimpeachable. If he says he can break HDCP, he can. Intel can claim that
the break is only theoretical and not practical, and nothing to worry
about, but I believe Niels is telling the truth. Kind of sounds like the
security world before full disclosure: without revealing details, the
affected companies simply lie and say there's nothing to worry about.
<http://www.securityfocus.com/news/236>
<http://www.wired.com/news/politics/0,1283,46091,00.html>
Ferguson's essay on the topic:
<http://www.macfergus.com/niels/dmca/index.html>
Yet another 802.11 wireless security flaw:
<http://www.eetimes.com/story/OEG20010803S0082>
<http://www.eyetap.org/~rguerra/toronto2001/rc4_ksaproc.pdf>
AirSnort is a new hacking tool for wireless networks. It grabs network
traffic and recovers the master encryption password, using some of the
vulnerabilities previously published. One interesting feature is that you
don't need to sniff all the packets in one sitting. Instead, you can
accumulate small increments over a long period of time to accomplish the
same goal. This can even be done with a program that runs when the laptop
is apparently asleep. That is, someone can quietly be sniffing your
network, even when they aren't actively using their computer. Maybe
they're just sitting in your lobby, waiting for a job interview, or
something equally innocuous.
<http://www.wired.com/news/wireless/0,1382,46187,00.html>
The next step is for someone to port it to a wireless PDA.
The crackdown on people with security knowledge continues. Brian West,
working at an ISP, was testing a banner advertisement at a local
newspaper. He inadvertently discovered a massive security hole in the
newspaper's Web software. After calling the newspaper to notify them, he
was arrested by the FBI. The more you read about this story, the weirder
it is.
<http://www.linuxfreak.org/post.php/08/17/2001/134.html>
<http://salon.com/tech/feature/2001/08/29/west/index.html>
FBI's side of the story:
<http://www.theregister.co.uk/content/4/21288.html>
A good essay supporting full disclosure:
<http://www.securityfocus.com/news/238>
Flash worms. Thirty seconds to infect the Internet. (This is an
improvement of the Warhol Worm idea I linked to last month, which can
infect the Internet in fifteen minutes.)
<http://www.silicondefense.com/flash/>
Really interesting paper on the likely future development of worms and viruses.
<http://www.crimelabs.net/docs/worm.html>
Secure your computer by filling it with concrete. It works, too.
<http://www.uoe.dk/csworld/security.html>
The "European Copyright Directive," basically the European equivalent of
the DMCA:
<http://www.eurorights.org/eudmca/CopyrightDirective.html>
One insurance company charges you more if you use Microsoft IIS:
<http://www.zdnet.com/intweek/stories/news/0,4164,2805929,00.html>
Under the DMCA, you're guilty until proven innocent.
<http://www.salon.com/tech/feature/2001/08/23/pirate/index.html>
For more than five years, McDonald's prize games have been rigged. This is
a great story about internal attacks and how damaging they can be.
<http://www.cnn.com/2001/LAW/08/21/monopoly.arrests/index.html>
The NSA has released a new version of Security Enhanced Linux, based on the
2.4.9 kernel.
<http://www.nsa.gov/selinux/>
Yet another eBook reader, this one by Microsoft, cracked:
<http://www.technologyreview.com/web/roush/roush083001.asp>
<http://www.zdnet.com/zdnn/stories/news/0,4586,2809412,00.html>
Interesting article about the liability of hacking victims. If, for
example, a brokerage company's network is hacked, preventing its customers
from making trades, can those customers sue the brokerage company?
<http://www.nytimes.com/2001/08/24/technology/24CYBERLAW.html>
See these more scholarly papers on the same topic:
<http://www.sidley.com/cyberlaw/features/liability.asp?print=yes>
<http://www.mazunetworks.com/radin-toc.html>
Yet another Code Red variant. I hope no one thinks these sorts of worms
are ever going to go away.
<http://www.zdnet.com/zdnn/stories/news/0,4586,2807674,00.html>
Flooz hack. People were buying Flooz with fake credit card numbers, and
then redeeming the currency at legitimate merchants:
<http://www.nytimes.com/2001/08/27/technology/ebusiness/27ECOMMERCE.html>
Really good opinion piece on the fallacy of perfect security and the
importance of risk management:
<http://www.osopinion.com/perl/story/13092.html>
The newly accused spy Brian Regan appears to have chosen a bad crypto
scheme to protect his secrets; the FBI was able to decrypt his files. My
guess is that he choose a good crypto algorithm, but that he used a
guessable key.
<http://www.securityfocus.com/news/245>
Here's an interesting risk. Increasingly, people are taking corporate PCs
for home use; for example, as compensation in lieu of pay from failing
dot-coms. And they're not changing the security settings. Hence,
computers that are configured for use behind a firewall are now being used
with no firewall.
<http://www.wired.com/news/culture/0,1284,46417,00.html>
My favorite quote: "Most [people] are totally unaware that computers can't
just be turned on and left to their own devices."
Russian hackers and organized crime:
<http://www.zdnet.com/zdnn/stories/news/0,4586,2784950,00.html>
Nasty new virus. This one is disguised as an e-mail warning from
Microsoft: they fake the e-mail "from" address. The text tries to convince
the user to run the executable, which then encrypts all executable files,
rendering them unusable.
<http://computerworld.com/nlt/0%2C3590%2CNAV65-663_STO63419_NLTSEC%2C00.html>
<http://www.cnn.com/2001/TECH/internet/09/03/virus.exe.idg/>
Security flaw found in the Gauntlet firewall. This is a big deal: someone
discovered a vulnerability in this brand of firewall that would allow a
hacker to take control of the box. Network Associates shouldn't be singled
out here; vulnerabilities have been discovered in other security
products. Again, I want to stress the importance of real-time monitoring
and defense in depth. It's the only way to get resilient security.
<http://www.securityfocus.com/news/248>
We're already seeing attacks against Windows XP, and the product hasn't
even been released yet.
<http://www.wired.com/news/infostructure/0,1377,46531,00.html>
Interesting Code Red footnote. It seems the FBI discovered a test version
of the worm in April, but decided not to warn people. Here's a good
example of how full disclosure could have helped.
<http://news.cnet.com/news/0-1003-200-7078908.html?tag=mn_hd>
Another worm, a "cure" for Code Red:
<http://www.secadministrator.com/articles/index.cfm?articleid=22381>
Online casinos hacked:
<http://dailynews.yahoo.com/h/nm/20010910/tc/tech_gambling_hacking_dc_1.html>
More IDSs vulnerable to Unicode attacks:
<http://www.theregister.co.uk/content/55/21573.html>
Son of DCMA. The SSSCA aims to make copyright protection manditory in all
computer hardware. This would be a disaster on all fronts: for privacy,
for security, and for the unfettered ability for technology to prosper.
<http://www.wired.com/news/politics/0,1283,46655,00.html>
<http://www.wired.com/news/politics/0,1283,46671,00.html>
Text of bill:
<http://www.politechbot.com/docs/hollings.090701.html>
This is something I predicted months ago:
<http://www.counterpane.com/crypto-gram-0105.html#3>
** *** ***** ******* *********** *************
Counterpane Internet Security News
Bruce Schneier submitted a brief in the Felten vs. RIAA case:
<http://www.eff.org/IP/DMCA/Felten_v_RIAA/20010813_schneier_decl.html>
Schneier is speaking at the ConSec conference in Austin, TX, on 24 September:
<http://www.consec.org/index.html>
Schneier is speaking at the ISSE Conference in London, on 26 September:
<http://www.eema.org/isse/>
Counterpane Internet Security, Inc. provides Managed Security Monitoring
services to large and medium enterprise networks. I formed the company
because I believe that expert human monitoring is the only way to provide
effective security, and that outsourcing this kind of service is the only
cost-effective way to provide it. This CERT document provides a different
perspective on the difficulty of an enterprise managing network security by
itself, and discusses some of the implications of Counterpane's
service. (Actually, the whole site is really useful, but the particular
document goes through a specific methodology for building a reasonable
enterprise-wide audit infrastructure. I believe that anyone who seriously
considers doing this themselves will quickly see the benefits -- both in
cost and quality -- of outsourcing to Counterpane.)
<http://www.cert.org/security-improvement/practices/p091.html>
<http://www.cert.org/security-improvement>
** *** ***** ******* *********** *************
The Doghouse: Shyfile
I guess we're back to shooting fish in a barrel; this is another snake-oil
alert. Shyfile claims to have a "self-decrypting JavaScript" that performs
6144-bit encryption on your e-mail to make it "hacker-proof." According to
their Web site (which has a bad-music soundtrack): "No generally known
algorithms such as AES, DES, IDEA or those used by encryption software
developers such as RSA, PGP or DEMCOM are integrated into ShyFile but only
its own (exclusively developed) enhanced TL6144D algorithm." They talk a
little bit about the algorithm, and it doesn't look too good.
There's even a contest. If you manage to decrypt a specific file, they'll
give you $1000. The contest expires in 2011; no word about the company.
<http://www.shyfile.net/>
** *** ***** ******* *********** *************
New Microsoft Root Certificate Program
"New root certificates are no longer available with Microsoft Internet
Explorer. Any new roots accepted by Microsoft are available to Windows XP
clients through Windows Update. When a user visits a secure Web site (that
is, by using HTTPS), reads a secure e-mail (that is, S/MIME), or downloads
an ActiveX control that uses a new root certificate, the Windows XP
certificate chain verification software checks the appropriate Windows
Update location and downloads the necessary root certificate. To the user,
the experience is seamless. The user does not see any security dialog
boxes or warnings. The download happens automatically, behind the scenes."
This is the kind of thing that worries me. What exactly is this
process. What happens when it fails. Why does everyone have to trust a
root certificate, just because Microsoft does. And if the user doesn't
see any security dialog boxes or warnings, the effects of any failure are
likely to be catastrophic.
For a while now I have talked about the differences between vulnerability
and risk. Something can be very vulnerable, but not risky because there is
so little at stake. As Microsoft continues to centralize security,
authentication, permissions, etc., risk rises dramatically because the
effect of a single vulnerability is magnified.
<http://www.microsoft.com/TechNet/itsolutions/security/news/rootcert.asp>
** *** ***** ******* *********** *************
Comments from Readers
From: Ryan Russell <ryan@securityfocus.com>
Subject: eEye and Code Red
It sounds an awful lot like you've been listening to people who keep trying
to convince us that eEye essentially released an exploit. They
didn't. Have you looked at their
post. <http://www.securityfocus.com/archive/1/191873>
Despite a comment in there about releasing a test exploit later, they
didn't. What they did do was give an academic discussion about how to
preserve shell code in a Unicode conversion environment, with no code
samples. Based on what you had to say about Sklyarov later in the same
Crypto-Gram, you would appear to be in support of such discussions.
Now take a look at the Code Red disassembly. Notice that the general
techniques discussed in the Bugtraq post were not used.
The hints that eEye gave didn't help the people who needed help writing
their overflows, and the people who understood what they were saying didn't
need help in the first place. The Code Red authors (the first version)
didn't need any help from eEye, that's for sure. The comments eEye made
about the exploit were little more than an in-joke for people who already
knew enough to figure it out themselves. This would be like you dropping
hints about S-box design; it wouldn't do a thing to help me replicate Twofish.
I'll repeat what I said when you made similar disparaging comments about
Mixter in the wake of the first DDoS attacks: "Would you rather have had
the month to act before Code Red was written, or not. Would you rather
have had the Microsoft patch, IDS rules, and details of the level of access
gained by the hole, or not?"
I agree with you about the complexity and speed of the Internet. We're
eventually going to get to the point where the next big worm is using 0-day
exploits that we haven't seen before. In fact, there was a blip on the
radar not too long ago; the precursor to Code Red, actually. It used an
unknown exploit. Fortunately, the author hadn't perfected his spreading
technique yet.
Given the choice, I'll take the month's warning, please. I'd rather have
eEye tell me than Mr. Worm.
> eEye is going too far.
They did free research for Microsoft, waited for Microsoft to release a
patch, didn't release an exploit, created a product that can help block
these kinds of attacks, wrote a free scanner to help people look for
vulnerable servers, stayed up all night on several occasions to disassemble
two worms for free (and the world was in serious need of those
disassemblies at that point) and then did what they could to help spread
the word to the media to try and get everyone to patch their servers. What
do you want from these guys?
From: Seth Arnold <sarnold@wirex.com>
Subject: eEye and Code Red
I am baffled why you badmouthed eEye in this month's crypto-gram. Marc
Maiffret at eEye posted a note to bugtraq, which should explain why eEye is
the last group that should be blamed for Code Red. (Mainly, that Code Red
used a different technique to exploit the problem than the one discussed by
eEye.)
Note especially that eEye coordinated a release with Microsoft, to ensure
that a patch was available *before* release. Note that Microsoft would not
have known about the .ida vulnerability without eEye alerting
Microsoft. (Read: we have no idea how long this problem has been known in
the hacker community.)
I am especially surprised at your position for one very basic reason: you
are underestimating the hacking community. You of all people should know
that keeping the details closely hidden will work for only so long before
attackers get all the details. If eEye could find this problem, chances
are good hundreds or even thousands of people could have found the problem
just as easily -- had any of them bothered to check.
From: "Ian Alexander" <ialexander@touchnetnw.com>
Subject: Microsoft and Code Red
It's funny that Culp at Microsoft told the Eastside Journal that only a few
MSN servers were affected. A friend of mine who works at Microsoft in
Redmond was telling me how their corporate network was toast for close to a
full day. It hit Microsoft a lot harder than they were comfortably
divulging to the newspaper.
From: Kee Hinckley <nazgul@somewhere.com>
Subject: Code Red and Personal Web Servers
One thing you don't mention about Code Red: it has triggered something
which I've been expecting for a while -- the segregation of the
Internet. A number of ISPs, mostly cable but some DSL, have blocked
incoming port 80 connections and show no signs of ever turning them back
on. Last week Mediaone/AT&T/@Home turned off incoming port 80 with no
warning whatsoever. To the press they claimed that consumers weren't
allowed to run Web servers anyway (something that is contradicted by their
Terms of Service, which specifically states that customers may run Web or
ftp servers at their own risk). To their customers they've said nothing at
all.
The fact of the matter is that keeping a server secure on the Internet is a
more than most individuals and small companies can handle. I spend a
significant amount of time every day tracking security lists, and while I
can justify it as part of my consulting business, most people can't. I
think what we're going to see is two types of Internet access. Companies
with IT staffs will be able to afford a raw Internet connection, but most
people are going to sit behind ISP firewalls that block virtually all
incoming connections and many outgoing ones. The Internet as the great
equalizer, where "anyone can publish what they want" is going to fade
away. Perhaps some ISPs will provide limited open connections for extra
cost, but the configuration and support issues are such that I think it's
unlikely.
From: "Enright, John" <johne@mail.webaccess.net>
Subject: Software Liability
There are several fundamental differences between tires and
software. Tires must hold up under a very limited, well-defined set of
tests. For a complicated software system, one simply can't test
everything, or all possible uses of the system; let alone try to think of
all ways a hacker might break through to create a security problem in one's
software. You yourself have said many times that hacking is a different
mindset; the security algorithm designer isn't generally able to break
his/her own system as well as others will be able to. It's difficult for
the builder to wear both hats. Most large software companies/corporations
have quality testing departments, but we can't expect them to catch everything.
The next very fundamental difference here is something you just pointed
out: tires are directly related to the safety of persons. Software
generally does not fall into this category, unless you start looking at the
medical field. Where lives may be at stake, you can bet that software is
very, very well tested (and there may indeed be accountability in this area).
You know, in a roundabout way, the market is more of a driving force to
make sure that a company's tires are of sufficient quality than the
accountability regulations themselves. Firestone/Bridgestone (and
similarly Ford) are undergoing very large financial burdens because of
this. It's not just the accounting for poor quality, it's the damage they
incur from such a public incident and loss of reputation. That will stick
with them for years to come.
So I guess we're comparing hardware safety regulations to software security
flaws. I don't see them as very similar. Maybe if you more clearly
defined what you mean by "accountability," I would better understand
you. It's absurd to demand accountability for every piece of software out
there. The lawyers would have a field day, and the software industry would
just stop. You'd have lawsuits on everything from carpal tunnel syndrome,
to eyestrain (from font size, etc.), to headaches, to security
breaches. Then software companies would start getting around this by
requiring that you sign some sort of waiver (more than that absurd and
annoying EULA everyone clicks through without reading now). Accountability
would have to be clearly defined. It's a bureaucratic nightmare.
From: Tim Kordas <timk@filanet.com>
Subject: Lockpicks and the DMCA
Lock picks are an excellent example. In all jurisdictions that I'm
familiar with, lock picking tools are on a list of "burglar's tools" (which
also includes things like vice-lock pliers, screwdrivers and
flashlights). Possession of burglar's tools alone isn't
illegal. Instructional manuals about how to pick locks aren't illegal (and
are clearly protected by the First Amendment), and are widely
available. Buying locks to pick isn't illegal either.
Possession of lock picking tools along with "criminal intent" to use them
in the commission of a crime is the illegal part; possession also can be
taken into account in sentencing on conviction (apparently they make the
possessor seem like a professional criminal). Presumably publishing
papers, and giving talks about the advantages or disadvantages of some
particular locking technology would be allowed (provided it didn't qualify
as libel).
The DMCA seems to grant an unprecedented set of previously nonexistent
"rights" to inventors of copyright technology. The provisions in the law
that make exceptions for "legitimate security research" specify that such
research cannot take the form of material which can be used to circumvent
the security measures! The easiest (if not best) way to show, in a
legitimate security research context, that any particular security measure
is in adequate is to provide at least a detailed description of a means to
circumvent the security.
There is no practical way of allowing legitimate security research to
proceed while disallowing invention of circumvention measures.
From: jjordan@unitechsys.com (Jason Jordan)
Subject: Adobe, Elcomsoft, and the DMCA
This is a scary quote from U.S. Attorney General John Ashcroft: "There are
many people of poor and evil motivations who are seeking to disrupt
business and government and exploit any vulnerabilities in the digital
universe."
<http://news.bbc.co.uk/low/english/sci/tech/newsid_1454000/1454489.stm>
There are many people of good intentions who are seeking to help business
and government by exploiting vulnerabilities in the digital
universe. Let's face it, if what Sklyarov did was "of poor and evil
motivations," he would have kept it quiet and not gone public with it. He
basically told Adobe that their e-book security was flawed. Instead of
Adobe saying something like, thanks, we'll improve it. They said,
government help! Instead of people taking the time and effort of making a
security system correctly, they make third-rate security systems and depend
upon laws to protect them from those who point out the flaws. If this was
the way security worked, we'd still have doorknobs on safes in banks, and
turning a doorknob would be illegal.
From: Jeff Tucker <jefft@wciatl.com>
Subject: Microsoft's Untrusted PGP Signatures
>This story is too weird for words. Microsoft
>adds PGP signatures at the bottom of its security
>bulletins, for verification. But if you try to
>verify the signatures, they fail.
This one I think you've got wrong. The problem is a ridiculous choice of
words chosen by PGP (now Network Associates), not Microsoft. This
particular issue (not specific to Microsoft) has been hashed out many times
on the PGP newsgroup.
The problem is that in PGP there is no central key issuing authority, as
you know. Everything is based on the "web of trust." Thus, I will
personally trust a few keys. If a person who I trust has signed your key,
I will trust you.
The problem arises because of the terminology PGP uses. If it decodes the
signature, but the key isn't trusted, PGP says that the key is "Invalid."
Obviously, this is ridiculous. A better choice of words would be
untrusted, unverified, whatever.
Microsoft can't do anything about this. PGP ships by default with NO keys
trusted. Keys can be manually trusted by individual users. Thus, if you
download the MS key from their secure Web site and figure it's the right
key, you really should sign the key with your own key. This will allow you
to trust the key, and the Invalid goes away. If you were to send me a
message with your own key today, I'd get the same thing: Invalid. Unless
one of the 20 keys I trust happened to sign yours.
This has lead to an amazing example, though, of how users can't be trusted
to understand cryptography and identity verification. At least one person
on the PGP newsgroup said that he knew how to get rid of all the
Invalids. He just routinely signs every key on his keyring and, voila, the
Invalid is gone. He might as well uninstall PGP, it would have the same
effect.
From: Jaap-Henk Hoepman <hoepman@cs.utwente.nl>
Subject: Council of Europe CyberCrime Treaty
Although the CoE treaty on cybercrime has improved, I would not call it
"kind of neat" yet. The treaty still makes it illegal to produce, sell,
etc. devices whose primary purpose is to circumvent protection schemes
without right. As you state elsewhere in the same issue of Crypto-Gram:
> Elcomsoft created and marketed a product that circumvented
> Adobe's product. This kind of software is often required in
> Russia, where people have a legal right to make personal backups.
There is no way really to tell whether the primary purpose of a device is
to make personal backups or illegal copies. So the CoE CyberCrime treaty
suffers from the same problem as the DMCA in this respect.
More worrisome is the fact that the treaty fails to consistently require
dual criminality as a condition for mutual assistance. This means that in
certain cases, states have to cooperate in a criminal investigation against
a citizen of that state even if that state does not consider the act under
investigation a criminal offence.
From: Adam Wagman <adam.wagman@cognex.com>
Subject: SirCam Worm
A worm that disseminates people's personal information was first described
in John Brunner's 1975 novel, "The Shockwave Rider." Brunner coined the
term "worm" in that novel.
From: Steve Andrews <sjandrews@chantilley.com>
Subject: Chantilley Data Security
I've read your recent comments in Crypto-Gram and have posted a reply:
<http://www.chantilley.com/Html/News12.htm>
In my reply I finish by inviting you to come to us in the UK and examine
our technology, or alternatively we can come to you. I do hope you will
accept my invitation so we can prove to you that our technology is sound
and that we do know what we're doing.
From: Eric Young <eay@pobox.com>
Subject: Computer/Telephony Integration
The telco security/privacy rot has already begun in Australia. One of the
long distance carriers -- AAPT <www.aapt.com.au> -- allows one to view
one's account information online, make payments, list all phone calls,
etc. Your login name and password. Account/customer number and
password/invoice number -- both of which happen to be consecutive lines on
your monthly paper statement.
What is worse is that this lets you sign up for ISP services, change all
your phone numbers over to AAPT, etc. It is also not possible to opt
out. Every AAPT subscriber is automatically on the Web. I sent some
e-mail complaining about this and asking how I could opt out. No response,
but there is now a mechanism to change the password to 5-8 characters, no
spaces, symbols or case sensitivity. When I originally complained, there
was no way to change the "password."
Total access for the effort of intercepting one monthly statement from the
mailbox. And any statement, no matter how old, will give a working
login/password.
From: John Giannandrea <jg@meer.net>
Subject: Computer/Telephony Integration
One thing you omitted from your "Phone Hacking: The Next Generation"
article is VoiceXML, a new markup language from W3C which will control all
manner of automated voice interactions: <http://www.w3.org/Voice/>. The
combination of VoiceXML and VoIP will be particularly vulnerable to
security attacks.
From: Operations@Digitour.com
Subject: Computer/Telephony Integration
A private telephone line is protected by the fourth amendment. The
Internet is a shared network, with no implied privacy. If these two are
mixed, do we all lose our fourth amendment rights on the telephone?
From: Markus Schumacher <ms@ito.tu-darmstadt.de>
Subject: Computer/Telephony Integration
We identified some vulnerabilities of a H.323 based IP telephony
scenario that incorporates gatekeepers and gateways. During our tests we
could:
* get control of an end-system
* perform DoS-Attacks using IP telephony signaling
* perform DoS-Attacks using general means
* attack the users' privacy
* attack infrastructure components
Those vulnerabilities are not due to limitations or shortcomings within
technology, architectures and signaling protocols. Again the main reasons
are severe flaws in design, implementation and policies of products.
You can find more details here:
<http://www.ito.tu-darmstadt.de/publs/papers/cms2001.pdf>
From: McNelis@aol.com
Subject: Copyright in History
Sorry to butt in to a conversation among IT professionals, but as a
professor of medieval British literature I have to take issue with Richard
Straub's interpretation of the history of publishing and copyright.
The argument that copyright was not needed by Chaucer or Shakespeare is
irrelevant in Chaucer's case, and misleading in Shakespeare's. Chaucer did
not write with any expectation that anyone, including himself, would ever
make money selling copies of his works -- except a manuscript shop which
might make one to order every now and then. Contrary to the assertion that
medieval books were all expensive and hand-copied by monks "in papal
states" (whatever that is supposed to imply), secular manuscript shops were
mass-producing books of wide interest as early as the 1100s in Paris, as
well as books needed for university students. Money was made by those who
made the physical books and sold them, but not by those who wrote them --
the critical problem from a modern point of view.
Shakespeare would have liked to have copyright protection, since in his
time plays were routinely stolen and performed without remuneration to the
company which created them. The only defense was to restrict each actor to
his own copy of the play he was performing, one which included only his own
part and no others. Garbled copies of plays nonetheless would be cobbled
together and shown in performances which were not sanctioned by, and
returned no box office receipts to, the actual authors.
Cervantes, author of "Don Quixote," watched helplessly as his book became
one of the most popular ever known in Europe without a dime coming his way
(from the original, that is). But since the 18th century, it has been
possible for people to make money as writers of drama or of novels -- only
because they receive payment for their work, passed on by booksellers,
publishers, or what have you under some sort of contractual
arrangement. The proposal that "rich citizens can afford to commission
works directly" sets the clock right back to Chaucer's time.
From: mercuri@gradient.cis.upenn.edu
Subject: Copyright and Radio
As a long-time member of the Musician's Union, I'd like to contradict your
statement that when radio was invented that folks didn't bemoan the fact
that people could listen to things for free. Actually, the Musician's
Union was highly concerned about copyright issues regarding radio
broadcasts, both for the performers and the composers of the works. What
they established (after lengthy negotiations) was a scheme whereby funds
were provided for airplays (in very small amounts per play), and placed
into pools that could be used to provide royalties and also pay for live
performances. In fact, when the phonograph became popular, and more
recently when electronic synthesizers became popular, musicians were
concerned again that live music would be replaced by recordings. In fact
some of this is true, but recorded music and synthesis has created other
opportunities for musicians, and expanded audience interest in
concerts. So although I totally agree with you that there needs to be a
better model for distribution of copyrighted works, perhaps even leaving it
for the market to decide might turn out to have beneficial side effects,
the example you provided about radio is not true to what actually happened
in the music industry in that regard.
** *** ***** ******* *********** *************
CRYPTO-GRAM is a free monthly newsletter providing summaries, analyses,
insights, and commentaries on computer security and cryptography.
To subscribe, visit <http://www.counterpane.com/crypto-gram.html> or send a
blank message to crypto-gram-subscribe@chaparraltree.com. To unsubscribe,
visit <http://www.counterpane.com/unsubform.html>. Back issues are
available on <http://www.counterpane.com>.
Please feel free to forward CRYPTO-GRAM to colleagues and friends who will
find it valuable. Permission is granted to reprint CRYPTO-GRAM, as long as
it is reprinted in its entirety.
CRYPTO-GRAM is written by Bruce Schneier. Schneier is founder and CTO of
Counterpane Internet Security Inc., the author of "Secrets and Lies" and
"Applied Cryptography," and an inventor of the Blowfish, Twofish, and
Yarrow algorithms. He served on the board of the International Association
for Cryptologic Research, EPIC, and VTW. He is a frequent writer and
lecturer on computer security and cryptography.
Counterpane Internet Security, Inc. is the world leader in Managed Security
Monitoring. Counterpane's expert security analysts protect networks for
Fortune 2000 companies world-wide.
<http://www.counterpane.com/>
Copyright (c) 2001 by Counterpane Internet Security, Inc.