The Risks Digest

The RISKS Digest

Forum on Risks to the Public in Computers and Related Systems

ACM Committee on Computers and Public Policy, Peter G. Neumann, moderator

Volume 14 Issue 54

Monday 26 April 1993

Contents

o CLIPPER CHIP
Dorothy Denning
Magnus Kempe
Jeffrey I. Schiller
Daren Stebner
Paul R. Coen
Tony Harminc
Padgett Peterson
A.E. Mossberg
David P. Reed
Jonathan Papai
Lauren Weinstein
o Info on RISKS (comp.risks)

Re: Responses to Clipper Chip Discussion in RISKS-14.53

Dorothy Denning <denning@cs.cosc.georgetown.edu>
Sat, 24 Apr 93 14:39:12 EDT
Don Alvarez wrote

  Now, I have enough faith in the NSA's skills to trust that you can't pass the
  message to the user without also passing the escrow field, but I'd like to
  see that in writing.

That is certainly the intent.

  First, who knows F? If the cop on the street knows F then it quickly
  becomes so widely known that everyone knows it and you might as well send
  the serial number in the clear.  Ergo only the escrow agents know F.

F is embedded in every Clipper Chip, but like other chip keys, unknown to the
people who use them.  Only law enforcement will have a decoder box that allows
the law enforcement field to be decrypted. Initially, there will be just one
box, and it will be operated by the FBI.

    Second, what information does law enforcement get from the
    escrow agents?  Ideally, law enforcement only gets K, the
    session key governing the conversation for which they have a
    court order.  Unfortunately, in order for law enforcement to
    get *only* K (and not U), the escrow agents must get together
    and secretly combine U1 and U2 so that they can unwrap K and
    give it to law enforcement.  But then the escrow agents would
    also know K, and they would be able to decrypt messages
    themselves.  That's just a very short step away from the
    omniscient big-brother which the multiple-escrower scheme was
    designed to prevent.  The escrow agents must not allowed to
    exchange keys with each other.  Ergo, law enforcement must
    assemble U itself in order to find out K.  This means that law
    enforcement now knows the key to unlock every session key ever
    used on this phone.

It is imperative that law enforcement get U.  If they are tapping a line,
there may be dozens of calls on that line per day.  It would be totally
impractical to have to go to the escrow agents to get the session key for each
call.  It would be impossible to do real-time decryption under that
constraint.

    Things are starting to look bad, but what about F?  We never
    really figured out what F was for, we just said that only the
    escrow agents know F.  Well, now we know what F is for.  Even
    if law enforcement knows U, they still can't read messages
    without court orders, because they can't obtain E[K;U] without
    knowing F, and only escrow knows F.

For the same reason as above, it is imperative that law enforcement be able to
decode the law enforcement field in order to obtain E[K; U] and then decrypt
this to get K.  It is completely impractical to go the escrow agents for each
conversation.

Steve Holzworth wrote:

    I don't claim to be up on crypto, but it seems to me that once
    the "authorized tap" has been performed once, the agency in
    question now has a copy of the desired keys. They can use these
    keys at ANY time in the future to decrypt communications
    between these two parties (without having to go through the

After a tap has been completed, government attorneys are required to notify
the subjects of the electronic surveillance.  At that point, the subjects are
certainly free to purchase a new device with a new chip, or perhaps the
vendors could simply replace the chip.

  I have one outstanding question that I haven't seen asked yet.  Is there a
  time component in the encryption key? Since wiretaps are presumably
  authorized for certain time periods with both start and end dates, it should
  not be possible to decrypt an illegally monitored message.

I am unaware of any time component.  Current wiretap laws protect against
this.  Evidence collected after the warrant has expired can be thrown out in
court.  In addition, it is illegal for the service provider to implement an
intercept after a warrant has expired.  With the new technologies, law
enforcers will be incapable of executing a tap without the assistance of the
service provider.

Lars-Henrik Eriksson has written

  If the algorithm was made public, any weaknesses would be discovered
  in time. If it is classified, weaknesses may never be known, or known
  only to the parties who have access to the classified information.

The NSA has a long record of success with crypto, far better than any
individual or organization in the public community.  In addition, there are
plans to bring in expert cryptographers to assess the algorithm.

Dave Weingart wrote:

    It appears ... that two encryption devices, when they initiate
    a session, must exchange keys in order to decrypt each others
    messages.  Great, but here's my question...what's to stop
    someone at machine B (who's talking to machine A) from
    "recording" the key from machine A when the session is
    started?  Since it appears that the key is constant for each
    chip, machine B can now _always_ decrypt machine A's messages.

The unit keys that are embedded in the chips are not exchanged.  Instead,
machines A and B negotiate a session key K that is used only for that
particular conversation.

Jim Sims wrote:

        Seems a whole lot easier to just catch the key K during the
        negotiation between the boxes....

It is possible for both ends to negotiate a session key K without transmitting
any secret information at all, including K.  One way of doing this is with a
public-key distribution method. The Diffie-Hellman method works as follows.
Machine A picks a secret value xa, and machine B picks xb.  A sends the public
value ya = g^xa mod p to B and B sends yb = g^xb mod p to A, where p is a huge
prime and g is a global constant.  Then A computes the session key K = yb^xa
mod p = g^xbxa mod p, and B computes K = ya^xb mod p = g^xaxb mod p.  Both K's
are the same.  An eavesdropper sees ya and yb, but since xa and xb are not
known, cannot compute K.  For more information about key exchange, see for
example my book "Cryptography and Data Security" or some other crypto text.

Dorothy Denning denning@cs.georgetown.edu

   [There were enough additional messages for several more issues,
   some raising points already covered by the above message.  I have
   somewhat arbitrarily selected a representative few for the rest of
   this issue, trying to avoid duplication where possible.  There may
   be another issue or two yet to come out of the existing backlog.
   PLEASE pardon some of the duplication.  It is virtually impossible
   for me to selectively choose a few nonoverlapping paragraphs from each
   message.  For those of you who wonder whether RISKS has been taken over
   by this discussion, there has been essentially no other topic of concern
   for the past week, although something may be brewing in the recent
   near-fatal aircraft autopilot failure attributed to software.  PGN]


Thoughts on the U.S. Encryption Proposal

Magnus Kempe <Magnus.Kempe@di.epfl.ch>
Fri, 23 Apr 93 09:47:25 +0200
Mr. President:

I am concerned that your proposal fails to address the following issues,
among others:

1.  Will American companies have to manufacture different lines of
products according to the market (US vs. foreign--assuming that they
wouldn't be allowed to export the Clipper Chip) ?

2.  Will foreign companies be excluded from the American market since
they won't have access to the Clipper Chip ?

3.  If the technology is ever sold to other countries (foreign
governments), how will the US be sure that no foreign government
develops the ability to tap American communications, and how will
international communications be secure (remember that the two escrow
"agencies" are to be located in the US, a seemingly unacceptable
proposition for non-American entities) ?

4.  What guarantee is there that other encryption schemes will not be
outlawed in the future?  In particular, what guarantee is there that
the right retained by the people to privacy will not be ever more
violated by the US government in the future?  Will use of other,
private, secret encryption become a crime?

5.  What guarantee is there that criminals will use the Clipper Chip,
knowing that the US government has the ability to decrypt their
communications?  Isn't it probable that such criminals will resort to
publically available encryption systems that they would know are secure
from the US government?

6.  Since the secrecy of the algorithm seems to be essential, how does
your proposal compare to the strength of currently available non-secret
algorithms?  It is well known in Computer Science that reliance on the
ignorance of an enemy party is the worst protection imaginable, since
it is a very weak link in the chain of safety.

7.  An American hero once said something to the effect that those who
are willing to trade liberty for safety deserve neither.  Don't you
think you are asking the American people to trade liberty (their right
to privacy) in exchange for an elusive safety (listening to suspected
criminals, if these criminals buy the government's chip)?

8.  How is it justifiable to ask the American people that they spend
money (for the Clipper Chip) in order to let the government listen on
them?  Given the current state of the economy, is it even possible?

9.  Finally, what constitutional article gives you the power to violate
the privacy of the American people?  Didn't you swear to uphold the US
constitution?

Sincerely,

Magnus Kempe

Software Engineering Lab, Swiss Federal Institute of Technology, DI-LGL /
EPFL, CH-1015 Lausanne, Switzerland +41-21 693 2580  Magnus.Kempe@di.epfl.ch


Re: Clipper Chip (Alvarez, RISKS-14.52)

Jeffrey I. Schiller <jis@mit.edu>
Fri, 23 Apr 93 21:53:09 -0400
   The summary isn't very specific, but it sounds like the message stream is
   transmitted first, followed by the escrow field.  Surely this can't be the
   case, because all one needs to do is hang up at the appropriate point in the
   call and prevent the transmission of the escrow field...

One way of solving this "problem" is to have the escrow block consist of:

  E[{E[K ; U],N,SHA[{E[K ; U],N}]} ; F]

Where SHA[] represents the Secure Hash Algorithm (in the Capstone
Chip), though frankly a CRC32 would probably do just as well. The
receiving Clipper can then decrypt each received law enforcement block
and verify that the SHA hash of the contents match the supplied hash.
If they fail, the chip ceases to perform any decryption operations on
behalf of the user.  Perhaps other information could be included in
the encryption under "F" to make it difficult to supply a law
enforcement block from another prior conversation. This could well be
a detail that was left out of the explanation given to Dorothy.

Btw. I will go on record as being opposed to this entire proposal. My
interpretation of the Constitution (and the Bill of Rights) on this issue is
that the 4th amendment (no search and seizure without a warrant) was intended
to place limitations on the right of the government to interfere with the
people. It is being twisted around by some to imply that we the people never
had the right to privacy in the first place, when a warrant was issued. As
others have mentioned here, I have a real concern that a VOLUNTARY program
today will turn into a mandatory program tomorrow. In my nightmares I envision
an agent of (pick your favorite large, secretive, government agency) telling
Congress:

"Yes Senator, but we gave the American Public privacy five year's ago with the
Clipper program. They still have that privacy today. However some miscreants
in our society... drug dealers and whatnot are still using non-Clipper
unbreakable encryption and we must put a stop to this..."

-Jeff


Clipper Chip Commentary (not suitable for the cynicism impaired)

Daren Stebner <STEBNERD@wl.aecl.ca>
22 Apr 1993 22:03:55 -0600 (CST)
Short Synopsis:

"We're from the government.  We're here to help...."

Long Synopsis:

   "No, no, no.  Don't use those regular old encryption devices; if everyone
uses them, then criminals will use them, too.  And if criminals use them, Big
Brother won't be able to protect you from all those nasty deals they make over
the phone.  It just makes the work of protecting you poor lost sheep that much
harder.
   "Here, Big Brother will make things all better, but in order for it to
work, you need to use this handy dandy NEW encryption algorithm that HE
designed.  He even put it on a chip to make it easy for you to use it in all
of your communication devices.  He even gave it a cute name -- the 'Clipper
Chip'.  See?!  Now, since Big Brother will know how to decode the messages,
He'll be able to listen in on the conversations of all of those yucky bad guys
so He can put them in jail and make things all better.
   "Could we listen in on your conversations too?  Well, yes, but we would
never dream of infringing on innocent peoples' rights to privacy.  That would
be a crime and Big Brother doesn't commit crimes.  He loves you all and would
never do anything to hurt you.  Most of all, he just wants to protect you from
those terrible, terrible, drug dealers.  That's all he would ever use the
Clipper Chip for.  Isn't that swell of Him?"

Daren Stebner  stebnerd@wl.aecl.ca


Clipper chip & databases

"Paul R. Coen" <PCOEN@DRUNIVAC.DREW.EDU>
23 Apr 1993 00:46:41 -0400 (EDT)
A few things have occurred to me as I've read the announcements and the
initial reactions.

Each chip contains and broadcasts the serial number.  To be frank, if they
don't use a lot of different family keys, then that number is going to be
rather public, rather quickly.

Legally, what is this going to do to wiretaps?  If I have recorded a
conversation, have I tapped it?  Or have I only tapped it when I decrypt it?
This could lead to recording lots of conversations, and then only getting the
keys and actually "tapping" once I get someone for something criminal.  A
change to the law or court ruling could establish this. For now, they need
permission up front (with exceptions, noted below), but I could see that
changing.

Doesn't the manufacturer need to give the keys to the escrow agencies? With the
serial number?  Am I missing something, or is there a RISK of the manufacturer
becoming compromised?

The other thing is that companies are going to be using this for international
calls.  And, calls in other countries.  I'm sure most of you have noticed that
agencies of the US Government tend not to follow the procedural niceties that
they at least have to give lip service to here when dealing with communications
elsewhere.  If the CIA or NSA decides they want a tap, can they get the keys
for a unit in another country without a court order, by saying "it's for
national security" three times and throwing a rock in the air?  Besides, the
NSA gets really itchy if they can't just monitor international calls whenever
they want.  So, what are they going to do -- record the call data and decrypt
it later if they need to?  They don't need specific authorization now to tap
and record international calls.  I can't see that changing.

What I read in one of the gov't documents implied that the A.G. was responsible
for defining some sort of reasonable authorization.  Why do I have a feeling
that "national security" is going to come up a lot?  Especially since there has
been an increase in cold-war style rhetoric in reference to *economic*
competition?  Regarding profitable technologies as national security concerns
for the sake of economic competition is a scary thing. And this chip plays
right into that desire for control.  As other people have pointed out, once
they've got your key, they've got your key.  If someone gets it for national
security reasons, then you might just be out of luck.

Paul Coen, Drew University Academic Computing   pcoen@drunivac.drew.edu


Clipper Chip

Tony Harminc <TONY@VM1.MCGILL.CA>
Fri, 23 Apr 93 00:29:38 EDT
Thoughts on the Clipper Chip:

1) One of the selling points of the Clipper chip is that US companies will be
able to use it to effect secure communication between their home offices and
branches in foreign countries.  In particular, it is implied that it is the
governments of those foreign countries that will be thwarted in their attempts
to listen in to the corporate secrets of America.

Now why would any "friendly" foreign government (e.g. Canada, France,
New Zealand) imaginably permit Clipper to be used on its territory
unless it too has access to the keys for "law enforcement purposes" ?

So if XYZ Corp. wants to talk in private with its French subsidiary XYZ
France, SA., the French government will want access to the escrow agents so
that it too can present a court order (according to French law, of course) and
be given XYZ's key, if it suspects wrongdoing on the part of XYZ France.  But
this clearly won't do.  The US escrow agents will presumably be subject to US
law and might be able to refuse a French court order on some US constitutional
grounds.  So the French will have to have their own pair of agents, and -
since there is no advance control of which chips will end up in France - these
French agents will have to have the complete list of all keys.  Now multiply
this by a dozen or so friendly countries, with an equal number of different
legal systems and constitutions...

2) Presumably the reason for keeping the algorithms secret is to prevent
competitive manufacture of chips (or software) that can communicate with
Clippers from being produced.  (Such competitors might somehow forget to send
their key lists to the escrow agents.)

I know almost nothing of the technology, but it seems far fetched to me that a
chip can be manufactured that *absolutely, positively* cannot be reverse
engineered, or at least satisfies something analogous to being computationally
infeasible to reverse engineer.

There was no mention of quantum effects, but I know of no other way to even
begin to make something that can't be examined with appropriate probes.  I
hope some hardware experts will say something on this topic.  Or is it that
the hardware design can be reverse engineered, but the algorithms themselves
are one-way encrypted ?

3) It is not clear to me how tapping of bidirectional communications
works.  If the police have a court order to tap the phone of suspected
criminal X, and they find that he is holding a Clipper-encrypted
conversation with previously unknown person Y, will they be able to
decrypt only what X says if they have only X's key from escrow ?
Or will they automatically apply for Y's key too, on the grounds
that he is an associate of X ?

Ordinary analogue phones (and networks) echo a small amount of the received
signal to the sender, but an encrypting phone will have digitized and
encrypted the signal before it gets echoed (even if there is a modem and
analogue circuit in the loop).

Tony Harminc, Apios Systems Toronto,  tony@vm1.mcgill.ca


Baltimore Clipper

407)826-1101 Fri, 23 Apr 93 08:46:34 -0400
With all of the sound and fury surrounding the announcement, I think a few
things have been missed. First, no-one has said that Clipper is going to
replace the STU-III or the Lockheed Encrypting Modem (just passed up an
opportunity to buy a couple for $25 - only 1200 baud) or all of the other
devices that exist. Elements of the Norton Utilities and PKZIP would also have
to be outlawed.  The "user selected table" in UUENCODE would be right out.

Second, prohibition of double encryption using Clipper as one stage would be
impossible to enforce through sheer mass. Comes under the same category
as strict enforcement of the speed limit.

So what is Clipper ? IMHO it will be a low-cost way to provide *reasonable*
protection for routine traffic that up until now has been unprotected.
Cellular telephony is the obvious first use to reduce the billion-dollar
fraud situation. Transmittal of medical records, legal records, credit
reports is another. Privacy laws state that these must be protected.
Clipper provides a legal remedy that has been lacking.

Telecommuting is another major problem for most companies who have been back
through fear of unautorized interception. Clipper will provide a "warm and
fuzzy" feeling with low-cost encrypting modems.

IMHO, Clipper must meet the "good enough" test. Clipper *will* meet the "good
enough" test because the designers are not stupid and it would be a major
*political* embarrassment should it prove to be easily broken, we just do not
know all of the facts yet.

Technically, I can make a guess and say that little of what we have seen
as yet is correct. For instance if the Message Key (K) is only 30 bits
long and the message is encrypted  E(M;K) as mentioned several times,
a massive attack with existing technology that is not particularly
expen$ive would yield a solution in under a minute. (Hint: check out
DSPs). Therefore the message is not encrypted E(M;K).

Further, there is a Family Key (F) and a Unit Key (U) in each chip. The
only link to these is the serial number therefore I *suspect* that the
serial number will be sent en clair as part of the header. It would be
possible to use only selected Family Keys and create E(s/n;F) such that
it would easily yield the s/n but why (more later).

So most likely we will have a message key that will be a predetermined
function of K,U,&F (FN1). The header might consist of the serial number,
followed by an encrypted function of K,U,& F (FN2) that is different from
(MK), then the message encrypted with MK.

Before engaging Clipper, both FN1 and FN2 would have to be exchanged.

On starting communication, the header would consist of the serial number,
followed by FN2, followed by E(M;FN1). To the receiving Clipper, the
serial number would act as a "wake up". Since the chip would have K,U, & F
it would then reverse FN2 and create FN1 and decode the message. A longer
header could accommodate conference calling. This removes any incentive
to spoof the s/n since the receiving chip will assume it is for someone
else. (Exercise is left to the student).

A "promiscuous" chip might be designed but it would have to have all
U and F keys (of course the total number issued is probably going to be
less than the number of ZIP codes so this *is* a danger point...).

The "key master" would hold one element say the s/n,U function, the
"gatekeeper" s/n,F. On proper application each would contribute its part
but the requesting authority would receive only FN1, not the means to
recreate it. Alternately, the requesting authority might receive a
duplicate Clipper chip but not the Keys - this protects the keys from
disclosure *and* provides physical traceability of the duplicate Clipper.
On completion of the tap authorization, the duplicate must be destroyed.

Certainly, there are still some things to be worked out but these are
technical details. IMHO we *need* a cheap means of "good enough" encryption,
something that can be built into modems and cell phones at a disposable
cost. We *need* a workable Clipper, and the government has more to lose
in providing a flawed product than we do.
                        Padgett


Re: Clipper Chip

a.e.mossberg <aem@symbi1.symbiosis.ahp.com>
23 Apr 1993 10:29:50 -0400
One thing (among several) that disturbs me about the Clipper Chip is the
release of the decryption key to law enforcement bodies, who have various
legal maneuverings to circumvent ever notifying the tappee, after the wiretap
has completed, that their key has been compromised.

The key, originally safeguarded by the key escrow bank, is now additionally
held by unknown and unaudited persons within law enforcement.

History has shown that the law enforcement community has not been above
using information gained during illegal wiretaps...  Once they have the
key, they have access forever to your conversations.

And, supposing that one is actually notified that they were being
wiretapped, and thus need a new key.  Will the agency responsible for
the tap pay any costs for obtaining new keys?

Further, the information packet provided by Mat Heyman's office fails to
address the question of using non-clipper chip encryption.  They call their
technology "more secure than many other voice encryption systems readily
available" -- admitting that more secure systems are available.  Why should
anyone believe they do not intend to make illegal competing methods of
encryption?

andrew mossberg, systems specialist, symbiosis corporation, miami, florida
33166-6202   (305) 597-4110   fax (305) 597-4002   aem@symbiosis.ahp.com


time, recording, and clipper

David P. Reed <reed@interval.com>
Fri, 23 Apr 93 10:52:26 PDT
In some of the recent comments on the RISKS of Clipper, it would seem that the
commentators are ignoring the results of including omnivorous recording.  It
is well known that it is possible to record a substantial amount of traffic
for archival storage and later analysis.  Such recording is not prevented by
any encryption scheme, whether keys are escrowed or not.

The resulting RISK of this assumption is that we may erode away the notion
that pure recording is a violation of privacy (today's wiretap law prevents
recording conversations without a warrant).  What could be wrong with law
enforcers or others recording everything, now that it is masked by encryption?
Surely this is NOT a violation of privacy because you can't read it.  I'd bet
that such recordings and archival would be authorized by most governmental
lawyers as NOT violating privacy on this basis.  But in fact, with key escrow,
someone on a fishing expedition could in principle recover all past history
from an archival database.

I think that the design of Clipper protocols is actually more suitable for
ex-post-facto decryption of recorded conversations, rather than for getting
access to not-yet-happening transactions.  So the notion that we might be
moving to a world where recording is real would make sense.

I also notice that the protocol for selecting conversation keys requires an
online conversation -- thus one cannot use Clipper for leaving secure voice
mail in a voice mail box, as far as I can tell.  Similarly, conference
calling, etc. need to be considered in the design of secure phones.  If
conference calling services (such as AT&T's service) require that the security
devices be turned off or compromised, there is a risk that users will be
deceived about their level of security.

My point, then, is that issues related to recording seem to have been
poorly considered in this proposal.  Perhaps I'm wrong, but I'd like to
understand this better, in the context of societal impacts that may result.

David P. Reed, Interval Research Corporation, 1801-C Page Mill Road
   Palo Alto, CA 94304


Who will use clipper?

Fri, 23 Apr 1993 11:18:42 PDT
An amazing amount of discussion and speculation has been generated by the
clipper chip announcement.  Special thanks to Dorothy Denning for her summary,
and kudos to Don Alvarez for asking good questions.  But I have yet to see
anyone ask (or answer) the obvious question:

Will the various government agencies be using the clipper chip?

If so, I heartily endorse the policy.  It guarantees that government messages
cannot be ultimately secret from the governed.  It seems to me that the private
sector will continue to use whatever encryption is appropriate for their needs,
(even should laws to the contrary be passed,) and that the only agencies
obliged live with the consequences are those of the government itself.

            --Bob

bebert.osbu_north@xerox.com -or- ebert@xsoft.xerox.com -or- (415) 813-7579
-or- XSoft/Xerox Corp. 3400 Hillview Ave. M/S PAHV203  Palo Alto, CA 94303


alt.privacy.clipper

Jonathan Papai <PAPAI@kcgl1.eng.ohio-state.edu>
Thu, 22 Apr 1993 23:34 ???
I guess someone should mention the existence of a new newsgroup
alt.privacy.clipper .  Might as well be me.

-Jon


privacy

Lauren Weinstein <lauren@cv.vortex.com>
Thu, 22 Apr 93 19:17 PDT
There is largely unique discussion of clipper going on over on the PRIVACY
Forum Digest.  For information regarding the PRIVACY Forum, please send the
exact line:

information privacy

as the BODY of a message to "privacy-request@cv.vortex.com"; you will receive
a response from an automated listserv system.

Please report problems with the web pages to the maintainer

Top