| Submission Procedure |
|
|
|
|||||||||||||||
Government, Cryptography, and the Right To Privacy
(HyperMedia Unit, University of Auckland, New Zealand jshearer@cs.auckland.ac.nz) (Computer Science Department, University of Auckland, New Zealand pgut01@cs.auckland.ac.nz) Abstract: The notion of a right to privacy of citizens in their communications is discussed in the context of an international movement by governments towards regulation of cryptography, and consideration of key forfeiture systems in national cryptography use. The authors argue that the right to privacy in communications networks is an issue of major importance, assuring freedom of the individual in national and global communications. Regulation and control of cryptography use on the Internet by national governments may lead to an imbalance in the citizen/government power relationship, with sequelae including unprecedented surveillance of citizens, disruption of international commerce due to lack of powerful cryptography (and lack of standardisation), human rights abuses by less democratic or non-democratic governments, and limiting of the political potential of an Internet global political system. Category: K.4.2 Social Issues (Computers and Society); K.5.2 Governmental Issues (Computers and Society); E.3 Data Encryption
1 Introduction
Cryptography use within the Internet has the potential to reorder citizen/government power relationships, a potential which is already attracting the close attention of national governments. Cryptography policy in the United States is the subject of low-level controversy, following the failure of the "Clipper" initiative, where the Government attempted to introduce a key forfeiture system. The EC has considered banning the public use of "strong" cryptography. The power shift initially appears to be due to uptake by Internet users of easy-to-use, freely available, effectively unbreakable cryptography. The result: completely private domestic and international communications, with the promise of follow-on "untraceable" digital cash. In response, governments are formulating policy with an unspoken subtext, which is a strong perceived interest in controlling cryptography use. The issues are major: economic advantage, national security, and law and order. How the balance of governmental controls and citizens' rights is resolved could have important political and economic consequences. The emerging scenario appears to follow on from a traditional "national" perception of cryptography as a weapon to be used in times of war, for secure communications by spies and the military. This paper will argue that not only has cryptography moved from the shadowed domain of confidential defence dealings into the public arena, but also that a raft of new issues are involved, for Page 113 example the novel clash between the interests of sovereign nations, and "global" interests of the increasingly politicised Internet community. In unravelling the complex issues involved in cryptography, it is helpful to look at three separate, though related, perspectives: those of the state, the market, and the citizens. In this way it is possible to weigh up the tradeoff of advantages and disadvantages to each group. So far, the gains and losses to national interests have been presented by law enforcement agencies as key matters in discussions of legislators in the US and in Europe. However issues across the board from human rights to small business security are potentially affected by the attitude taken by national legislators to cryptography. Cryptography is central to questions about how free the citizens of the future are going to be under the conditions of the future Internet. Development in the United States appears to be heading towards a push by a few large companies to a one-way channel to consumers of information as a commodity: home shopping, movies, packaged information for areas such as education and health [ Johnson 1995]. The Internet as marketplace requires cryptography only as a means of keeping commercial transactions safe: it is in the area of political discussion and other public forum functions that the importance of cryptography as a social issue becomes apparent. The Internet offers piecemeal information, and may lose out to large commercial information providers because of this. However, it already offers a politicised environment where newsgroups and lobby groups actively inform the Internet "community". This function may be seen as an important new "public good" which possesses potential for a global public forum. Its point of vulnerability in political terms is that it operates under a system where internal control is extended only to technological structures and necessary bureaucratic measures. Internet users are, in fact, highly vulnerable to the placing of national "security" measures which may impact heavily on individual privacy.Privacy of communications is not considered to be a human "right" in most countries as, for example, it is argued in international forums literacy should be a human "right". The Japanese constitution is unusual in guaranteeing citizens privacy of communications (Article 21 of the Japanese Constitution states in part "No censorship shall be maintained, nor shall the secrecy of any means of communication be violated", making Japan one of the few countries with constitutional guarantees of privacy). The authors are arguing that privacy of communications should be assumed to be a "right" of citizens, unless governments can produce instances in which national interest may rationally be found to override the "right to privacy" of citizens. We would argue that citizens should appropriately take a keen interest in any arguments put forward by governments to make exceptions to the right to privacy. The technology of the Internet is outstripping the capability of ethics developed on a global basis, for example, to create a global viewpoint on the balance between privacy needs and national and economic interests. The outcome is the potential for infringement of personal privacy on an unprecedented scale, a phenomenon which should not be viewed with complacency in even the most "free" societies. Strenuous efforts put in by national security Page 114 services to increase surveillance over citizens have been documented in the US, a situation which has led to concern in a society which considers itself to be one of the freest in the world. After all, if people are going to communicate and conduct business every day on the Internet, much as they used to over the fence or at work down the road, they will wish to do this without eavesdropping by the neighbourhood busybody. In the years since Orwell's "1984" was published, the term "big brother", representing a government which wishes to maintain total surveillance and control over citizens, has become a modern cliche. This book created a sense of outrage when it was published, but in an era when surveillance techniques are beginning to approach those of Orwell's imagination, a desensitised population is failing to protect a traditional, and vital personal freedom. The implications of loss of communications privacy are major concerns for human rights, for example in countries where oppressive political regimes find an interest in maintaining the fiction that all subjects agree with their views. The takeover of the Internet, juxtaposed with a political regime intent on using covert surveillance measures to monitor dissent and track the activities of people considered suspect, could smother political dissent, a goal which has been aspired to many times historically, but never before achieved. Most local area networks may be perceived as "spy networks" in which each node watches all the information flowing over a shared wire and picks out only those messages intended for it [McLeod 1993]. It takes only a minor modification (such as putting an ethernet card into promiscuous mode) to allow one machine on the network to watch all information for all machines on the network. When used in a firewall situation, a 486 PC can handle packet filtering at full T1 bandwidth [BorderWare 1995] meaning that a single PC discreetly inserted anywhere along the long link tying a geographically isolated country like New Zealand to the rest of the world can undetectably monitor all Internet traffic for the entire country. Given the low cost involved - a one-off investment of a few thousand dollars - and the scope of the possible return on this investment, it is clearly a temptation for governments to perform this kind of surveillance. The potential problem is not isolated to oppressive regimes, but is likely to appear in a different (illegal) form in traditional democracies, where citizens traditionally place a high value on privacy. The implications are wider than someone being annoyed that their financial status has been leaked to a wider audience than they would like, or that they appear on a consumers mailing list they don't want to be on. The implications for the less secure democracies are considerably more sinister.
Data security without "strong" cryptography is problematic, and the
United States, playing from a position of strength in the area of
cryptography development, is providing an example of a country which
is strongly favouring its perceived interests as a national
government, but not entirely at the expense of refusing to listen to
the concerns of its citizens. Thus in the United States cryptography
is classed as munitions and hardware or software implementations are
not allowed to be exported [ Page 115 [Root 1991] [Department of State 1992] [Relyea 1994]. However, "strong" cryptography and Public Key Cryptography (PKC) is available for use within the US, and in a very restricted manner, for communications to US interests outside the country. With the development of public awareness of and debate on government surveillance measures in the US, we see the emergence of a more mature understanding of the significance of cryptography. However, despite significant victories for the "electronic civil rights" lobby, the issue remains unresolved, with a central issue of a key forfeiture system (this term is used instead of the alternative "key escrow" since it more accurately describes the act of involuntarily surrenduring privacy safeguards to an external agency) still being pushed by US Government agencies. The covert regulation of encryption by governments has generally been more comprehensive and more successful than any overt regulation. This covert regulation takes the form of patent secrecy orders on cryptographic patents, the cutting off of funding for research into promising areas of encryption technology, the discouragment of standardisation attempts for encryption systems, and documented harassment of providers of encryption technology aimed at ensuring they stay in line with government thinking. Americans, with a traditionally strong interest in protecting individual freedom, are confidently attempting to create a balance between the rights of the citizens to privacy and the control of terrorists, drug dealers, and so on. How the US decides these issues may be a useful lesson to other nations. It is by no means certain the American people will tolerate the imposition of a key forfeiture system, and in the absence of this policy in the US it may be viewed as a very risky venture in terms of democratic politics for any other nation. That it is being seriously considered in legislatures as far apart as Australia and Europe indicates the extent of misunderstanding of the cryptography phenomenon (an example of this was illustrated by a recent call in the German parliament for a ban on encryption devices targeted at the level of encryption technology which existed before the second world war [Schroeder 1995]). The two methods which have been suggested to date, key forfeiture and weak encryption, are deeply flawed. Both schemes appear to negate most of the protection gained by encryption. The first, key forfeiture, requires trusted agencies who will hold the keys. To date no acceptable agents have been found. The main reason for this is the somewhat dismal record of existing government agencies which hold records on citizens. For example, the General Accounting Office (GAO) has stated that the FBI's computerized National Criminal Information Center (NCIC), established in 1967, is "routinely" used for unauthorized purposes by federal, state and local agencies [McPartlin 1993] [Madsen 1993a]. In San Jose, in the US, it is claimed police officers have sold information on individuals obtained from the mammoth Criminal Justice Information System system for $25 per report [Mercury News 1993]. The situation is little better outside the US. In the UK, many banks allow tellers to access any customer account; the information is then sold to anyone willing to pay for it [Luck and Burns 1994]. Page 116 In Australia, the New South Wales Independent Commission Against Corruption (ICAC), conducting an investigation into allegations of widespread unauthorised access to personal data, found that information from a variety of State and Commonwealth government sources and the private sector had been freely and regularly sold and exchanged for many years. The organisations involved comprised a virtual who's who of Australian banks and insurance companies, as well as Australian Customs, Australia Post, the Department of Immigration, the Department of Motor Transport, the Department of Social Security, the Police, Telecom Australia, and various local councils and other government bodies [Clarke 1992]. The commission concluded that ".. a massive illicit trade in government information ... has been conducted with apparent disregard for privacy considerations, and a disturbing indifference to concepts of integrity and propriety ... Laws and regulations designed to protect confidentiality have been ignored ... [Even where criminal sanctions existed] information ... has been freely traded". In light of such reports, public confidence in key foreiture systems is likely to be, quite properly, low - as one writer was moved to comment, "trusting the government with your privacy is like having a peeping tom install your window blinds" [Barlow 1994]. An alternative key forfeiture proposal involving non-government agencies has run into similar problems. For example Bankers Trust, one of the organizations in favour of key forfeiture and who would like to become commercial key escrow agents, have recently been accused of massive fraud and corruption - with 6,500 tapes and 300,000 pages of written material as evidence [Business Week 1995]. Even taking the ultimate step of using the military as escrow agents is problematic because of the long history of cryptographic equipment and keys - exactly the material which is meant to be kept secret in key escrow - being leaked to outsiders, often for trifling rewards [Allen and Polmar 1988] [Polmar and Allen 1989] [Blum 1987] [Barron 1987]. The second of the two methods, weak encryption, is equally problematic. The main objection to this means of encryption regulation is that any encryption capable of being broken by the government is equally capable of being broken by any other government, or by large corporations, or organized crime, or a drug cartel, or even a student with access to some spare computing time. The largest publicly admitted application of computing power to cryptanalysis was the factorization of RSA-129, a part of the 1977 RSA Challenge [Atkins et al 1994]. This effort consumed 5000 MIPS-years of computing power over a period of 8 months (it is estimated that the same result could be obtained in about a quarter of the time using a somewhat better algorithm [Lenstra and Lenstra 1993]). With a little added financial incentive, specialised hardware can be obtained to speed up the task (for example an add-on card for AT-class PC capable of giving it a multiprecision math performance somewhat better than a four-processor Cray XMP cost about $4,500 in mid 1992 [Dubner and Dubner 1992]). The RSA-129 effort, carried out on a purely volunteer basis mostly by students, is more than many governments would be willing, or even able, to commit towards breaking an encryption scheme. This problem is further complicated by the steady advance of available computing power. Encryption which is rated as "weak" today will be classed as "laughable" in a few years time when Page 117 more powerful computers become available (it has been postulated that the easiest way to break an encryption scheme requiring the investment of 30 years of computing time is to do nothing for 29 years, then break it in 1 year using the computers available at that time). Since secrets worth encrypting will often need to be kept confidential for years, even decades, it seems futile to try and protect them with a scheme which will be broken within the useful lifetime of the secret they are meant to keep. Banks and similar organisations already send huge amounts of data in encrypted form over electronic networks. Providing the ability to decrypt such data is an open ticket to commit financial fraud, and both weak encryption and key forfeiture encryption open electronic commerce systems to fraud. The same applies to electronic payment systems, where the use of this form of encryption is roughly the same as giving an attacker a blank cheque which can't be stopped and which has no withdrawal limit - once a secret key is compromised, there is no way to "un-compromise" it, leading to few limits on potential fraud. Similarly, industrial espionage is already big business and will only get bigger, and with bigger rewards come bigger temptations, so that attackers with the ability to decrypt sensitive communications and stored data are a very real threat and security liability for companies. The potential for damage is not limited only to financial and business information, but extends also to areas such as medical and personal data. For example, UK doctors guard their patients medical records with some care, and recently refused to put them online in unencrypted form as was called for in a national plan. Weak or key-forfeiture encryption would allow medical records to be accessed without the permission or knowledge of both doctors and patients, raising serious privacy concerns.
What are the reasons for the chaotic international situation regarding
cryptography? Democracy, shown to be a relatively fragile institution with a history
of only a few centuries, may not have the legal and political
structures in place to cope with the massive changes to information
transfer which will result from the Internet becoming the new
universal means of conducting human affairs, business, and personal,
and political communications. The cryptography issue is a primary case
in point. Citizens are required to cope with a new perception of
cryptography, formerly the domain of defence egg-heads and highly
classified usage in times of war or national danger. The risk is that
public opinion, an important part of democratic structures, may not
"operate" in the area of cryptography, because of public ignorance or
a tradition of entrusting "military" matters to governments. Yet, if
the governments move to compose legislation protecting "national
security interests", or promoting further convenience or power in
their own operation, they may trip over major emotional values of
their citizenry: those pertaining to individual freedom. These values
may not be articulated by a majority of the population, but will show
themselves when the invisible "line" is crossed. A conflict already
exists between users of cryptography demanding complete freedom of use, and the
pragmatic following of economic and security agendas by governments. In terms of the argument relating to cryptography, governments should
not perceive citizens as merely the geographic collection of people
under their governmental control. While devolving state functions,
limiting the power of labour unions, and so on may have modified a
number of traditional pathways for influences on state policy, other
loyalties and ties have developed. One of these is the feeling of
"community" Internet users have towards the Internet. To limit
effective use of the Internet by restricting access to cryptographic
techniques, or by blocking the development of global standards,
governments risk collectively offending users of the global Internet,
on the grounds of loss of individual privacy and data security, and on grounds of inhibiting global commerce. The issue, as yet something of a "sleeper" outside the Internet, is
likely to develop as the Internet community gains some control over
some of the more high-profile problems such as hacking and material
considered harmful. Cryptography regulation may increasingly be seen
by national governments as a means to control the new medium, as the
Internet takes on its own identity in the area of mass communications,
discussion forums and information systems, and digital commerce and
digital cash move past the experimental phase. At this point, the
effects of government regulation of cryptography use will become
evident to the citizens, as a major control on their personal freedoms
and privacy. Such regulation could also have the flow-on effect of
limiting the potential of the Internet as the means for a global
political movement. How the Internet community itself perceives the
potential uses of cryptography is likely to affect how strenuously
cryptography is defended by the community, as a means of achieving
individual privacy, establishing a digital marketplace, and creating
new political pathways via the Internet.
The overt regulation of cryptography in the US is done through the
classing of cryptography as munitions. Interestingly enough, the
Internet itself was created as a munition, a "reliable means of
transmission during events of unreliability", more commonly known as a
nuclear war. Export of encryption technology from the US is
occasionally allowed for large financial institutions which can prove
it will only be used for data authentication purposes, or if the
encryption is deliberately crippled to make it easy to break. Although
the US government claims that "anyone can apply to export encryption
technology", noone has ever been allowed to export anything other than
very weak encryption systems (it is generally accepted that if any
encryption technology is approved for export by the US government then
it can't be any good. However, the converse is not true - unexportable
crypto isn't necessarily strong). An example of weakened, exportable encryption technology is Netscape
Communications' World-wide Web browser, which generates a unique
128-bit session key for a transaction which is then used with a fast
encryption algorithm known as RC4 to protect the rest of the
transaction. To comply with US export restrictions, Netscape transmits
88 of the 128 key bits "in the clear" along with the message, so that
only 40 bits of the session key are actually kept secret. In July
1995, a French student used spare processing time on around 120
computers to break an encryption challenge posted to the Internet, in
8 days [ Another attack shortly afterwards took 32 hours, although it was estimated that a technical glitch caused it to take twice as long as it should have (a different attack, which takes advantage of an implementation flaw in the Netscape client software rather than the weakness of the encryption, takes about 1 minute on a cheap workstation). Another type of attack, which tests multiple sets of keys at once, is even faster [Collins 1995]. These successful attempts demonstrate the future security risk to businesses outside the US using weakened encryption. However the weak encryption does make the software acceptable to the governments of some countries such as France which normally ban encryption [DISSI 1995]. A number of attempts have been made to challenge the US export restrictions, both through attempts to change the existing laws via new legislation, and in legal challenges based on a claim that the ITAR contravenes the First Amendment to the Constitution, which guarantees freedom of speech [Kruh 1986b]. So far, all of these attempts have failed, on grounds of national security interests.
Like the US, France defines encryption hardware and software as
munitions. The "decret du 18 avril 1939" defines eight categories of
arms and munitions; the "decret 73-364 du 12 mars 1973" specifies that
cryptography equipment belongs to the second category; the "decret
86-250 du 18 fev 1986" extends the definition of cryptography
equipment to include software; and the "loi 90-1170 du 29 decembre
1990" states that export or use of encryption must be approved by the
French government. A documented effect of the French ban on the use of
encryption has been the increased ability of French intelligence
agencies to perform industrial espionage on non-French companies
operating in France. Foreign companies operating in France are
required to register keys for any encryption systems they use "for
reasons of national security". The head of the French DGSE (Direction
Generale de la Securite Exterieure) secret service has publicly stated this organisation helped French companies acquire over a billion dollars
worth of business deals from foreign competitors in this way [ Admittedly, the US (and for that matter a great many other countries) are little better than the French in this respect. For example, in the late 1970's the CIA set up an "Office of Intelligence Liaison" within the US Department of Commerce to pass information obtained by US intelligence agencies operating listening stations in other countries on to US companies [CBC 1994]. There are two such stations operating in New Zealand, one at Tangimoana north of Foxton and one at Waihopai near Blenheim. Similar listening stations also operate in other countries, with their (mis)use for industrial espionage being admitted by US intelligence agencies [Markt & Technik 1994] or reported in the press [Reuters 1994]. Recently, there has been a scramble by US companies to take advantage of US intelligence capabilities for industrial and economic espionage purposes under a variety of euphemistic labels such as "strategic information acquisition" [Brod 1995]. The CIA, after restructuring in the 1980's, is now itself entering the field, providing their services not only to US government officials but also to organisations such as the Department of Agriculture and the Federal Aviation Administration (FAA) [CIA 1994].
In contrast to the long-standing French restrictions, the Russian ban
on use of encryption was only recently introduced [ Page 121
In July 1995, the Australian Government tested the waters of
encryption regulation in a curious paper which, although presented by
the Assistant Director for Security Management of the Australian
Attorney-General's Department in a session attended by representatives
from the Australian Defence Signals Directorate (DSD) and the UK
Government Communications Headquarters (GCHQ), was marked as being
"the views of the author and not necessarily representing the views of
the Australian Government" [
The German government also appears to be moving towards restricting
privacy technology. On 4th May 1995 the German cabinet passed the
Fernmeldeanlagen Überwachungs-Verordnung, or telecommunications
surveillance bill, which requires that almost all communications
carriers provide a standardized interface to allow monitoring by
government agencies. This covers telephones, cellphones, ISDN, and
computer networks. Additional information such as call setup
information and data to allow tracking of cellphone users within cells
has to be made available. Finally, the creation of a universal
database listing the users of these services is required [ Page 122
The use of encryption has been considered by various political parties
in the UK, with most of them being in favour of it. The British labour
party, after initially coming down against encryption on the advice of
various governmental security advisers, changed their policy after
feedback from Internet users so that their current position is that
"attempts to control the use of encryption technology are wrong in
principle, unworkable in practice, and damaging to the long-term
economic value of the information networks. Furthermore, the rate of
change of technology and the ease with which ideas or computer
software can be disseminated over the Internet and other networks make
technical solutions unworkable. Adequate controls can be put in place
based around current laws covering search and seizure and the
disclosure of information. It is not necessary to criminalise a large
section of the network-using public to control the activities of a
very small minority of law-breakers" [
Another example of partial regulation of encryption was South Africa,
which in the mid 1980's passed a law that civilians could only use
encryption if they gave the South African army not only full details
of the algorithms and protocols, but also copies of all keys in
use. The banks sent a message to Pretoria to say that they welcomed
the idea of handing over the keys to their ATM's to the army, and that
whenever any of them were out of balance at the end of the day they
would send the bill to the government (banks use encryption primarily
for legal reasons - the key used to derive the PIN from an account
number and miscellaneous other information is kept in secure hardware,
so no bank employee can ever find out a customer's PIN. Since the
banks'security procedures are always completely foolproof and above
reproach, the only way the balance could be out is if a dishonest
member of the army had misused the keys held by the army to help
themselves to some cash. Therefore it was only proper that the
government foot the bill for this). After a long silence, Pretoria gave an assurance that the banks could
go on breaking the law and nothing would happen to them [ Page 123
A few governments have taken halfway steps towards regulating
encryption. For example, the Norwegian government has introduced its
own encryption standard called NSK, a secret stream cipher algorithm
in a tamperproof chip which can only be used under tightly controlled
conditions [ Other countries have also worked on tackling the "problem" of encryption technology. The Dutch government looked at banning encryption in 1994, but backed down rapidly over a storm of protest [Remijn 1994]. More seriously, a number of totalitarian states such as China (which recently required that all Internet users register with the police), Iran and Iraq are known to place heavy penalties on the unauthorised use of cryptography. In general countries follow recent directives which replace the older COCOM rules restricting export of cryptographic hardware and software to a number of countries including the former Soviet bloc, to ones covering a much smaller list of countries such as Libya and Iraq (the members of this list change with time). An example is the Austrian law on foreign trade [AHG 1995] which follows the equivalent EU directive [EU 1995] almost verbatim.
The US is retaining US developed encryption systems in its own hands,
for "national interest" reasons. In economic terms, the US Government
is evidently mindful of the penalties which may follow export and
international use of encryption and encrypted commercial transactions
on the Internet. The issue is overshadowed by the policy position of
the United States, an unrivalled superpower with an economic system
burdened by state overspending and high national debt. The
vulnerability of the US economy in the event of loss of tax revenue,
for example, has caused the issue of cryptography to become charged
with multiple implications. The United States has expressed commitment to the future of the
information superhighway, of which the Internet is a major growth
area. However,in the area of cryptography policy it is not surprising
that the US Government is tending to regard the major problems which
it is faced with as problems that will be solved by the US according
to its national needs. In international relations terms, this means
the US is appearing to place its national security interests and
economic interests before considerations such as the best future
development of the Internet or the economic well-being of other nations. The time gained by retaining export
controls on cryptography export may be essential time for the US to
deal with the regulatory implications of digital commerce and to
develop an effective working digital cash model, which may then be
imposed on the Internet as a de facto standard. Policymakers are
already showing signs of developing control mechanisms using software
patents, and export bans on certain types of encryption. This "you
want it, but you can't have it" scenario is not likely to advance the
development of the Internet - partly because of time frames, and
partly because the US is creating an encryption environment,
intentionally or not, which dictates to users of the Internet how
encryption will be used. Internet users of cryptography are advocating
the dropping of controls over publicly developed cryptography, as
Internet development is clearly penalised by the lack of distinction
made between the various issues by US government agencies, and the
corresponding lack of clear-cut issues presented for public debate.
One of the main impediments to the widespread use of encryption
technology today is the lack of any well-recognised international
standards guaranteeing interoperability between different
implementations. The sole internationally-standardised encryption
algorithm, DEA-1 [ A similar battle occurred over the attempt to standardise triple-DES encryption in the US. DES had long been recognised as being past its prime [OTA 1987] [Smid and Branstead 1988] [Federal Register 1992], and a new triple-DES standard was seen as an attempt to prolong the life of the cipher into the next century [X9 1994]. Triple-DES is popular because it can be easily incorporated into existing systems using DES, is based on standards and procedures familiar to most users, and can be made backwards-compatible with single DES with an appropriate choice of keys. Page 125 The NSA circulated a document among the members of the ANSI X9 standards committee [Rainville 1994] urging a negative vote on the proposal based mostly on the fact that triple-DES is "counter to national security and economic concerns", a curious claim since the reasons for X9 working on the triple-DES standard were to provide better protection for financial information than that afforded by single DES. A few months later,work on the triple-DES standard was approved [CDT 1995a], providing a major setback for the NSA who were now faced with the threat of a standardised encryption algorithm providing more strength than the Skipjack algorithm used in Clipper, but without Clipper's key forfeiture mechanism. The availability of triple-DES implementations received a further boost shortly afterwards with the announcement by AT&T and VLSI Technologies that they were developing new data security products based on triple-DES. Triple-DES hardware had already been openly available outside the US for several years [Cryptech 1989] [CEI 1992]. However while trying to restrict the civilian use of encryption on the Internet, the US government has recognised the need for encryption by fielding its own encryption system for the transmission of classified documents, voice data, and video teleconferencing - by the US military only [Aviation Week 1995]. Government interference in encryption work is not confined to the US. The Sesame project, a clone of MIT's Kerberos designed to provide authentication but not privacy, had its DES implementation replaced with a 64-bit XOR (and even that, it turned out, wasn't implemented properly) at the insistence of the EU's Senior Officials' Group (Information Security) (SOGIS), which consists primarily of signals intelligence managers. Researchers on another project, RIPE [den Boer et al 1992] were paid to devise a hash function but forbidden to work on any form of encryption [Anderson 1995] When it comes to public-key encryption, government intervention in standardisation attempts have also been quite successful [Price 1989]. The result has been an almost complete absence of international standards covering the form and use of public-key encryption systems, and of encryption algorithms which can be efficiently implemented in software. The effect of this is that cryptographic privacy protection, where it exists, is of an extremely ad-hoc nature.
Frequently the issue of privacy protection through encryption is
ignored entirely because nothing is easily available to perform the
task. One situation in which this is glaringly apparent is in the
cellular telephone industry. Analog cellular phones have no privacy
protection mechanism, making it very easy to intercept conversations.
Although the most widely-publicised means of interception are radio
scanners, these present a needle-in-a-haystack approach to monitoring
and make it almost impossible to target a specific phone or
conversation. The best cellular phone interception device is another
cellular phone. Details on converting cellphones to allow interception
of calls are often available from the phone manufacturers [ Page 126 The cellular phone industries response to this problem was to lobby the US Congress into passing the Electronic Communications Privacy Act [ECPA 1988], which requires people to pretend not to listen to the parts of the spectrum which contain cellphone traffic. Amusingly, some older television receivers with UHF tuning can tune the frequencies used by cellular phones, making it possible to break the law by tuning a television to the wrong channel (cellular phones operate on the frequencies formerly occupied by UHF television channels 70-83, which can be tuned by TV sets made in the 1970's or earlier). Had low cost encryption technology been widely available then the cellular phone industry might have provided real security to their customers rather than the "security" provided by the ECPA, as well as avoiding some of the US $1.5 million/day losses incurred due to cellphone fraud [Wilder and Violino 1995]. The encryption used by GSM cellphones is another example of national interests taking precedence over genuine security. When GSM was being developed during the 1980's there was intense debate among the NATO intelligence agencies over whether the encryption used should be weak or strong. Countries like West Germany, which shared a long border with an eastern neighbour known for its strong cryptanalytic skills, wanted strong encryption. Countries like the UK wanted weak encryption. The result was an algorithm called A5, which has been characterised by UK cryptographer Ross Anderson as "not much good" [Anderson 1994b]. A simple brute-force attack requires searching 2 40 keycombinations (the same number as the Netscape attack), with much faster attacks being possible. Interestingly, A5's low upper limit on the number of possible keys would seem to meet the US government requirements for weak exportable encryption. Attacks faster than the basic brute-force one are also possible, and one such attack was to be presented by Dr Simon Shepherd at an IEE colloquium in London on 3rd June 1994. However the talk was cancelled at the last minute by GCHQ. A chip to break A5 is currently being designed for an MSc thesis [Anderson 1994c]. However, even A5 was regarded as being too strong for export outside Europe. The result was a watered-down version called A5X, which was even easier to break [Lloyd 1993]. Countries like Australia, which managed to obtain cellphones employing A5 encryption, had to carry out multimillion dollar retrofits to communcations equipment to allow government monitoring of cellphone conversations [Lagan and Davies 1993] (the high cost of converting existing cellular phone networks into cellular monitoring networks has led at least one GSM vendor to claim that the cost of breaking GSM security itself was US$56M, this being the cost of the cellular network conversion as carried out in the Netherlands). Another alternative when governments find it impossible to monitor cellular communications is simply to ban them altogether [Griffin 1995].
Attempts to discourage research into encryption have occurred almost
continuously for nearly two decades. In July 1977, NSA employee Joseph
Meyer wrote to people planning to attend an upcoming symposium on
cryptography that participation might be unlawful [ In 1978, the NSA tried to block a patent on the Phasorphone, a cheap, simple telephone scrambler, but the secrecy order was revoked after an outcry in the media [Gilbert 1981] [LA Times 1994]. In the same year they tried to silence a University of Wisconsin computer scientist who had invented an encryption device [Kruh 1986a]. The chancellor of the university denounced the NSA for obstructing academic freedom, and the agency backed off [Markoff 1992]. In 1979, NSA Director Bobby Ray Inman, in an address which came to be known as his "the sky is falling" speech, called for encryption to fall under the same "born classified" doctrine which covers nuclear weapons research under the Atomic Energy Act of 1954 [Levy 1993]. In 1981 the American Council on Education (ACE), under pressure from the NSA, formed the Public Cryptography Study Group, which somewhat reluctantly recommended a trial scheme for the voluntary submission of crypto papers to the NSA as an alternative to the NSA's proposals that either the NSA monitor published technical information and recommend criminal prosecution if it was seen to threaten national security, or that the submission of technical papers to the NSA for prepublication approval be made mandatory, with publication without NSA approval being a criminal act [ACE 1981]. This scheme was again stopped by a media outcry [CFP 1994]. In 1982 the NSA tried to re-classify large amounts of previously public and declassified information used by James Bamford in his book on the NSA [Bamford 1982]. In 1984 National Security Decision Directive (NSDD) 145 gave the NSA authority over all government encryption and computer security development. In the same year the American Association for the Advancement of Science commissioned a series of ten study papers to investigate the ways in which secrecy and openness influence the conduct of scientific research [AAAS 1984]. In 1985 NSA Director for Communications Security Walter Deeley called for government regulation of encryption, stating that "it is time to put the genie back in the bottle for the good of society" [Deeley 1985]. In 1986 there was an attempt to extend NSDD-145 to cover the private sector. In the same year the NSA proposed a system in which they would provide all encryption equipment and keys for use in the US. This equipment would use NSA-designed classified algorithms with the special property that only certain types of keys would provide strong encryption, making it necessary to obtain all encryption keys from the NSA [Kolata 1986]. Opposition to this scheme was not long in appearing [Deavours 1986]. In 1989 the NSA attempted to stop dissemination of Ralph Merkle's "Khufu" encryption algorithm [Merkle 1991], one of the first very fast, secure software encryption algorithms (one of the authors has in his posession a yellowed printout of the Khufu paper, containing a hand-written note explaining it's "publication without NSA approval"). In the 1980's the National Science Foundation had a clause in its rules for graduate student fellowships requiring fellows to inform the NSF of any discovery "likely to influence national security". In June 1994, NSA agents visited Page 128 Jim Bidozs, president of RSA Data Security Inc, to talk about Clipper and RSADSI's competing products. After about two hours of discussions, one of the agents threatened to kill Bidzos because of the work his company was doing [Bank 1994]. A senior agency official later apologized for the incident, stating that it was not agency policy to make death threats. Just how little things have changed in the encryption debate is shown by a dissenting opinion from a member of the ACE Public Cryptography Study Group, which raises a number of basic points which are just as valid now as they were fifteen years ago when the report was originally published [Davida 1981]. In addition to discouraging work on encryption products, the NSA has also worked to block any software which might somehow work with other encryption products. For example, in May 1995 the NSA requested that the capability to interface with external encryption software be removed from the NCSA WWW server [NCSA 1995]. Although the server contains no encryption code, the mere possibility that it might be hooked in at a later date were enough to attract the attention of the NSA. Similar problems have also beset other attempts at providing internationally-available encryption products by adding encryption capabilities outside the US [Walker 1994].
The Electronic Frontier Foundation was formed to champion the civil
rights of computer users and to roll back a perceived attempt by the
various arms of the US government to control what happens within the
Net. The electronic civil rights movement has expanded to take in
other issues, of cryptography, and wiretapping. The movement is
questioning the need for extended state surveillance of private
computer and telephone communications. For example when the FBI filed
notice in the Federal Register in October 1995 requesting an increase
by 1998 to one thousand times the number of taps officially carried
out by the FBI in 1993, requiring that phone companies and other
service providers build enough surveillance capacity into their
systems that 1.5 million phone lines, or 1% of all lines in the US
could be simultaneously wiretapped, calls isolated, and their contents
forwarded to the FBI [ David Chaum, a pioneer of untraceable digital cash transaction technology on the Internet, places a high value on the privacy achieved by secure cryptography: "The choice between keeping information in the hands of individuals or of organisations is being made each time any government or business decides to automate another set of transactions. In one direction lies unprecedented scrutiny and control of peoples' lives, in the other, secure parity between individuals and Page 129 organisations. The shape of society in the next century may depend on which approach predominates" [Chaum 1992]. Many of the problems associated with hacking may be prevented by use of encryption of information, which effectively sets boundaries around private, as opposed to public domains. Cryptography is used because of the risks of loss of security caused by hackers and criminals. Obviously, cryptography may be used by criminals or terrorists to formulate plans for crime or to actually carry it out. However, it could be argued much of current US policy making is the product of a particular mindset in regard to economics and security, and that the stated fear of US officials about encrypted computer crime may have limited justification.
By late 1995, the US Clipper Chip initiative was generally
acknowledged to have failed. The reasons for this have been covered
exhaustively elsewhere, with two very in-depth discussions being
[ Another problem with Clipper was the discovery by an AT&T researcher that the key forfeiture mechanism built into Clipper devices could be bypassed without too much difficulty [Blaze 1994] [Markoff 1994a] [Quittner 1994]. Clipper messages can also be "forged" without a need to know the encryption key [Lomas 1994]. A final nail in the coffin was the release to the Electronic Privacy Information Centre in August 1995 of declassified FBI files which revealed plans to outlaw any encryption other than Clipper [FBI 1993] [FBI Undated a] [FBI Undated b]. Although heavily censored, these documents still contain enough information to show that at the same time as the US government was publicly promising to keep Clipper as a voluntary standard [Harris 1994], it was secretly planning to outlaw any encryption which the government couldn't decrypt in real-time... unless that encryption was used by the government to protect its own secrets. These documents added weight to claims by anti-Clipper groups that Clipper would only serve its purpose if all other encryption were outlawed. On 6 September 1995, the US government unveiled a new proposed crypto policy at a Key Escrow Issues Meeting [NIST 1995]. This policy gave 10 criteria which government-approved encryption systems would be required to follow, in return for making the resulting system exportable. The response to this proposal by Page 130 representatives of several of the largest hardware manufacturers and software publishers and various public interest groups was almost uniformly negative [CDT 1995b]. Clipper itself failed to meet many of the requirements, including (at least) No. 1, No. 2, No. 5, and No. 6. The main problem with the proposal, quickly dubbed "Clipper II", was that it required both weakened encryption through the use of short keys, and key forfeiture. Several conference attendees claimed there is no legitimate purpose served by limiting the key length on a system for which the government already holds the keys. The short-key requirement was seen as an attempt to preserve an extra-legal alternative to legitimate access via the escrow agents, one which sidesteps any need for a search warrant or other judicial approval. Several of the other criteria (such as No. 2, which prohibits multiple encryption) seem to reinforce this, making it possible for interested US government agencies (and well-equipped outsiders) to decrypt communications even without the escrowed key. It was also postulated that, since the 64-bit key is too small even for today, the whole Clipper battle could be re-fought in a few years time once attacks such as the current problems with Netscape's 40-bit keys are extended to 56-bit or 64-bit keys. Another problem was criterion No. 6, the requirement of non-interoperability with non-escrowed products, seen as yet another attempt to coerce key forfeiture without actually admitting to it directly. As with Clipper, it appears this requirement was designed to ensure that incompatible government-approved encryption would eventually flood out any competing systems. Yet another problem was that, as with Clipper, it was seen as unlikely that foreign governments would embrace a system which was conducive to US spying [EPIC 1995], especially in the light of evidence that the US had already in the past sold software with trapdoors in it to foreign governments [Madsen 1993b]. Finally, liability for compromised key databases was seen as a problem by a number of companies, with a Shell Oil representative stating that "the US government cannot cover Shells liabilities" in the case of compromised keys protecting data such as geologic information and market strategies, which were worth staggering amounts of money. At present it looks like Clipper II may go the way of the original Clipper [Markoff 1995b] (Almost every non-government speaker at the Key Escrow Issues meeting prefaced their remarks with some form of "Do not assume our presence here is an endorsement, because it is not...". One speaker suggested having a t-shirt made up with this on it to save everyone having to mention it at the start of their presentation). The government representatives said that they heard the comments, but would proceed anyway. In an interesting reversal of the usual pattern of events, a group of 37 companies said they would formulate their own crypto policy and present it to the US government within six months [Corcoran 1995]. Page 131
In the face of opposition to any form of government regulation of
encryption and related invasion of privacy, it is interesting to
speculate on the direction future attempts at encryption regulation
will take. The most difficult problem is proving that regulation is necessary. To
date, governmental attempts at demonstrating a "problem" have been
fairly unsuccessful, consisting of trotting out the so-called "Four
Horsemen of the Infocalypse" (porn, pedophiles, terrorists, and drug
dealers) as justification for encryption bans and increased
surveillance powers over communications. These claims have been
frequently challenged. For example, in a recent debate over Clipper,
FBI Special Operations agent Jim Kallstrom attempted to justify
Clipper by claiming it would help protect children from being
kidnapped to make snuff movies [ Even the claims of the need for greatly enhanced wiretap capabilities are somewhat questionable. For example in 1992 of the 39 states which have wiretap statutes, 17 reported zero taps that year; of the federal jurisdictions, 44 reported fewer than 10 taps for the year, including 19 who reported one tap and 36 who reported zero. The largest number of taps was reported by New York police, with 197 wiretaps installed [Wiretap Report 1992]. When FBI Director Louis Freeh lobbied Congress for the 1994 Communications Assistance for Law Enforcement Act (CALEA, better known as the "Wiretap Access Bill"), he cited FBI statistics claiming only 1,157 federal, state, and local electronic surveillance orders for all of 1993. To put these numbers into perspective, the FCC estimated that in 1993 the US had approximately 500 million phones covering 150 million phone numbers. Even the FBI itself seems unaware of any real problems in conducting wiretaps caused by encryption technology [Markoff 1994c]. The Wiretap Access Bill, S.2375, was passed with the unanimous consent of the senate, without any floor debate or reading of the bill, after a number of senators received a personal visit from FBI director Louis Freeh in the days before the vote [Bunker 1994] [Matthews 1994]. Page 132 A view often advanced of the move towards increased surveillance and encryption regulation is that, with the end of the Cold War, a number of signals and intelligence agencies are experiencing difficulty justifying enormous budgets in the face of cutbacks in other areas of the ecomony (the US government spends more money - US$28 billion - on intelligence than it does on housing or education [Toledo 1995]. This budget was increased by 5% for 1996, a 1.3% increase over and above the requested amount). It appears that the various intelligence agencies may be moving from concerns over national security to concerns over job security, requiring a new mission to justify their budgets [Markoff 1994b]. For example, the Canadian Communications Security Establishment was recently criticised for carrying out economic espionage on Mexico during NAFTA talks and on Korea to facilitate the sale of Canadian nuclear reactors, with a former CSE employee admitting to CTV news that the CSE shifted its focus after the cold war from spying on the Russians to spying on Canadas allies and trading partners in order to acquire trade secrets [CP 1995]. In the face of strict encryption regulation or even the unlikely scenario of a complete ban on the use of any form of encryption, there still remains a means of communication which cannot be banned because it cannot be detected: steganography, the art of hiding one message inside another. Such techniques have been in use to keep communications secret for centuries, with the first known use being by the astronomer Aryabhata in around 500 AD, who used a technique which mapped numbers to letters which could yield cipher words which were meaningful text [Kak 1988]. More recently, the British War Office devised a steganographic protocol which allowed soldiers in WWII prisoner-of-war camps to communicate information in their letters despite intense scrutiny by prison camp guards [Rabson 1990]. To date the most common use of steganographic techniques is in the game of bridge, where its use to allow bridge partners to communicate secret information in direct view of their opponents has caused a certain amount of controversy [Winkler 1980a] [Winkler 1980b]. Due to the nature of the communications channel, the amount of information which can be transmitted via steganography is normally very limited (the WWII cipher would, for example, require an entire letter to conceal a few short phrases about enemy troop movements). However with the advent of essentially free computer communications this restriction on size is lifted - an expansion of a hundred to one for a simple message is no longer seen as a major problem, since at worst it will require a few seconds longer to transmit the "carrier" message, with the messy details of complex en- and decoding being taken care of by the computer. Communications by computer-aided steganography can take place through virtually any form of overt communication, with messages being hidden inside sound files, pictures, or text (typical methods involve inserting message bits into the least significant bits of graphical images or sound samples, or making minute changes in letter spacings in text). Because the hidden messages can be made arbitrarily difficult to detect by making them arbitrarily close to the expected characteristics of the carrier message, the result is an undetectable means of communicating in secret - a form of encryption which cannot be banned or outlawed. Software which implements various steganographic techniques is already Page 133 freely available, and has the potential to become widespread if more conventional means of securing data are outlawed.
The development of trading of goods and services in the Internet may
drive the use of cryptography, and to some extent, force the hand of
governments as to its use. As the market develops, larger sums of
money will be circulated, and, presumably, criminal activity will
upscale accordingly. Secure cryptography may be perceived as necessary
to protect transactions, in the way that secure cryptographic
protection for banks is already seen as valuable. Thus, the market may
cause cryptography to lose the mystique of its traditional defence
role, and it may be seen by consumers as another product of the
information age that they wish to buy. It is likely that as consumers
become more acquainted with the product, they will demand better
services. With encryption programs like PGP already in wide circulation outside
the US, this development is likely to be rapid, and if the US holds on
to its isolationist policy in regard to cryptography for too long, it
may face the major economic risk of another nation producing
high-quality cryptographic software, and setting a new standard
outside the US. Such software is already being produced in countries
outside the US. However, in the face of the US market dominance, and
refusal to deregulate in the area of cryptography, it is likely the
market will remain fragmented and without definitive standards for the
forseeable future. It is likely that important attempts will be made
by large multinational companies moving into the market to establish
the technical standard of "adequate" cryptographic security, and to
look towards the establishment of global standards.
The Internet backbone was set up with United States Government money
and support, and the principle of an information superhighway is
supported by the US Government. However, there is a strong impulse in
the US and other countries to claw back political control over the
Internet. Particularly problematic is the unprecedented scope of
surveillance methods. These measures, being put in place possibly
before the American people fully grasp the significance of them, may
become the status quo, and difficult to shift in the future. However,
in the area of cryptography, the US is facing a quiet rebellion on a
number of fronts. One is the domestic resistance to the key forfeiture
proposals and legislation which electronic civil rights activists
believe will infringe individual privacy and freedom of Americans. The
recent strong lobbying efforts by the Internet community in 1995 in
respect of the Exon Communications Decency Act (where the Internet
community believed legislation to control offensive material would
damage the Net), and the resulting turnaround between the Senate
passing the Communications Decency Act legislation and the Congress passing the Cox-Wylie Amendment, (a more
low-key and practical approach to the problem) would indicate the
Internet community in America is rapidly learning to use its
teeth. Another advance is the pragmatic arming of other countries with
the weapons of future commerce, such as cryptography, securing of
electronic communications against piracy and damage, and Internet
access and literacy. These factors are likely to proceed to the point
where the US technological supremacy may be under threat, and
deregulation of cryptography will become unavoidable. Economic and
defence adjustments would then have to be made. However, it is
possible these may be more to the perceptions of Americans, rather
than to the possibility that due to secure encrypted communications,
the American economy may suffer disastrous damage, taxes will suddenly
not be paid, the war against drugs will be lost completely, and
bombers will run amok. Governments of sovereign nations will each be in the position of
deciding the trade- off between perceptions of security problems,
protection of civil rights, and economic advantage. The cryptography
issue may be seen as an issue of the relationship between government
and citizens, with the Internet and cryptographic technology having
the potential to substantially change the relationship. With complete
privacy of transactions and the ability to dodge many traditional
bureaucratic checks, a cryptography-based economy and society could
cause governments to become shut out of many business and social
transactions, unless people voluntarily allowed them in. The authors
would argue that the new environment established by the Internet
rightly demands a rethink of the social contract between governments
and citizens, and that this contract must be viewed in its totality,
as a contract involving issues of personal freedom and privacy, as
well as governance. A power imbalance achieved by governments as a
result of vastly increased ability to perform surveillance on
citizens, may be seen as breaking the "collective enterprise" [ The alternative is that an unprecedented, and undesirable, amount of power may come to reside in the government of countries, if key forfeiture cryptography schemes are introduced internationally. With its strong civil rights movement, the US Internet community has been well-placed to fight initiatives such as the Clipper Chip. That the Clipper Chip idea went as far as it did, is an indicator of how the rights of individuals in less democratic countries could be compromised if encryption "trapdoors" are built into national cryptographic systems, or if key forfeiture cryptographic systems were established and misused by national security agencies. The potential for human rights violations resulting from governments being able to gather "evidence" against dissidents on an unprecedented scale, is a major problem of new technology of surveillance being allied with cryptography regulation. In general, cryptography policy may develop from commercial needs, privacy needs, and the need to protect societies. This last category should be generated by the Page 135 Internet itself. No one country can do it without imposing significant penalties. The potential of an ethical community of Internet users to control criminal activity, for example, is a good question for the Internet community to ponder. Many of the concerns of the Clipper architects are demonstrably real. Issues of encrypted criminal or terrorist transactions, and drug money laundering (with associated uncontrolled casino activity on the Internet) are issues that the Internet community should rightly address. However, these issues should be separated from the cryptography debate, and addressed as political issues for internet community members, rather than as problems addressed only by national law enforcement or defence agencies. If an issue thrown up by the debate is the relationship between governments and citizens, it is a worthy subject for the Internet community to study in terms of planning its own political future. If the Internet remains a politically anarchic system, it risks losing its community forum and its potential future as a global open information system to repression by national governments. In the climate of governments moving towards regulation to limit use of cryptography and to establish key forfeiture systems, it makes sense to look at the possibility of an Internet political movement as a protective device. Just as the US Association for Computing Machinery is calling for a major public study on the uses of encryption on the Internet, the Internet itself should be creating a major study field of this critical issue, and associated issues of criminal conduct using encryption. Existing Net organisations like the Web Society could have a major part in this. A logical issue for the Internet community to address is that of effective cryptography standards for the conduct of business and personal communications. Public research into cryptography should be open, and the products of that research freely distributable without restriction. A point to keep in focus when considering regulating security aspects of communications media like the networks used daily: a new technique for cryptography may appear in any moment which would foil any efforts to monitor or police the exchange of encrypted data. The potential of steganography, for example, sends a warning to governments which attempt to censor Internet communications through cryptography legislation. To demonstrate the difficulty in regulating (or even detecting) this means of communication, messages using each of the three steganography techniques mentioned above have been embedded in this paper. Acknowledgements The authors would like to thank Professor Bob Doran for helpful advice during the writing of this paper. Page 136
ID<1994Apr15.124341.20420@news.research.ptt.nl>, posted to the
nlnet.cryptographie newsgroup.
Page 146
|
|||||||||||||||||
