May 1997

1. In March 1997, the Department of Trade and Industry (DTI) issued a public consultation paper setting out detailed proposals for legislation on the licensing of trusted third party services for the provision of encryption services. This paper set out the response of The Federation of the Electronics Industry (FEI) to the consultation document
2. The UK is now the fourth largest IT, electronics and communications (ITEC) consumer in the world, spending some £48 billion a year on IT, telecommunications, software and computer services. It also has the fifth largest ITEC industry, with a value of some £43 billion or 6.7% of total UK GDP.
3. FEI is the UK’s leading trade association for the information technology, electronics, communications, defence, office equipment and office furniture industries. FEI currently represents around 70% of the UK electronics sector by revenue, from global leaders to specialised, high technology SME’s, including world class players in their field.
4. We note that the DTI proposals:

• are intended to promote privacy and trust in an Information Society, and are only in part concerned with the special issue of law enforcement access to encrypted messages; and
  
• relate only to the regulation of trust services and will not mandate or prohibit any aspect of the use of products containing cryptographic functionality.

1. We welcome the publication of the consultative paper and the opportunity to respond. The FEI would also welcome the opportunity to discuss, at first hand with appropriate officials, any issues arising therefrom.
2. The DTI proposals represent the first government attempt anywhere in the world to bring together in a single licensing regime the general issue of the regulation of trust services and the special issue of access to the decryption keys of encrypted intercepts. Although we have critical comments (set out in the following sections) on this and some of the details of the consultation paper, we support the broad thrust of its thinking, and the notion of the licensing of Trust Service Providers where a service is provided to the public. However it will be essential for any regulatory system that addresses these problems to be technology neutral: We consider there is a danger that overly-specific laws promoting a specific approach to cryptography could enshrine current technology and restrict innovative computing schemes. We would also be concerned if there was an adverse impact the development of the Information Society in the UK. We are convinced that the UK needs to be a leading player in the establishment of a Trust Services Infrastructure (TSI) for Europe and the World. In particular, we see a major role for Britain in pioneering a policy and regulatory approach that could be replicated in due course in most developed economies world-wide.
3. The benefit for the UK, Europe and the World will be a Global Information Society based firmly on privacy and trust.

 

4. A TSI will enable and facilitate:

• the agreement and enforcement of contracts between parties who only ever "meet" in cyberspace;
  
• new ways of doing business, and a new level of effectiveness in the delivery of government services;
  
• the rapid formation and development of business partnerships and "virtual" organisations, combining and re-combining the strengths and skills of different companies to meet fast-changing competitive challenges;

• highly effective routes to market for new and existing products, extending levels of customer reach and range; and
  
• personal privacy and business confidentiality across electronic networks - while at the same time allowing for the safe restoration of decryption keys which are lost, damaged or required for audit or law enforcement purposes.

• Technically inaccurate. Trust Service Providers will offer a whole range of services but it is unlikely that the direct encryption of data will be among the more important services offered. "Cryptographic services" is certainly more accurate than "encryption services", and relates better to the notion of "cryptographic method" defined in the recent OECD Guidelines.. "Trust services" gives an even better indication of what is on offer.
  
• Misleading. The use of the term "encryption services" puts the overall emphasis of the consultation paper on confidentiality and warranted access rather than on building a TSI. The pervasive use of the term throughout the paper may explain many of the reactions that DTI has received from various groups.

• two strangers, having authenticated themselves using a trusted third party, then need a trusted third party in order to establish a secret encryption key;
  
• TTP's know the private value of the public keys in certificates they have signed;
  
• a TTP's knowledge of the value of a user's private key-exchange key will meet the minimum functional requirements set out in Annex E of the consultation paper; and
  
• knowledge by a law enforcement agent of the value of a user's private confidentiality key is the only, or preferred, method of realising what the OECD Guidelines call " lawful access to plain text, or cryptographic keys, of encrypted data".

None of these axioms is valid.

1. In an interactive scenario (for example, between a web browser and web server) - and, increasingly, interactive scenarios rather than store-and-forward implementations will be the dominant paradigm - the two ends of the transaction can exchange an encryption key (by Diffie-Hellman negotiation, for example) without needing to call on the services of a third party. Such two-party negotiation is convenient, widely used, and secure and can be expected to be the method of choice in almost all commercial applications.
2. In a store-and-forward implementation (for example, e-mail) one method of key establishment depends on the receiver's public key certificate being published in a directory. There is no technical requirement for the certificate authority signing the certificate, or for the directory service provider, to know the private value of any public key appearing in the certificate. Indeed, many product implementations take care to ensure that the private key is securely held so that it can never be shared or otherwise revealed. A smart card system, for example, will tend to include mechanisms to generate the private key randomly on the card in such a way that it cannot be later extracted into the outside world.
3. Even if service providers somehow knew the values of the private keys of their clients, and were thus able to give them up against a warrant, then the conditions of Annex E of the consultation paper would still not be fully met. This is because the transmitted data would include the message decryption key wrapped only with the public key of the recipient. Requirement 6 (... both incoming and outgoing communications ...) would not be met - only incoming communications would be accessible under warrant. Service providers would be unable to meet the full requirements of law enforcement agencies - and they would be unable, either, to provide an adequate key restoration service to their subscribers.
4. If key restoration meets the requirements of Annex E by being "balanced" then it always depends on the sender's specially-engineered encryptor including in the transmitted stream some extra protocol elements whose sole purpose is key restoration. This is true for key recovery schemes (such as IBM SecureWay Key Recovery) and for key escrow schemes (such as the Jefferies/Mitchell/Walker - or "Royal Holloway" - scheme).
5. The legislation establishing the regulatory regime and any new laws on warranted access each need to reflect this fact - as do any new UK export control rules. This can be done by recognising that a full-function key restoration capability depends on a framework which includes conformant product implementations and the availability of a whole range of trust services - not just the holding of keys by service providers. It should also be recognised that key escrow implementations and key recovery implementations are both able to meet the key restoration requirements of users and the warranted access requirements of law enforcers.
6. There also needs to be an understanding of the services that service providers will be able to offer law enforcement agencies. For example, the following five "law enforcement services", offered by service providers to law enforcement agencies, all have different characteristics:

• LAW ENFORCEMENT SERVICE 1: provision of a key restoration key.
  
• LAW ENFORCEMENT SERVICE 2: provision of a secret triply shared by the sender, recipient and service provider.
  
• LAW ENFORCEMENT SERVICE 3: provision of a secret value used to wrap a message decryption key.
  
• LAW ENFORCEMENT SERVICE 4: provision of a message decryption key.
  
• LAW ENFORCEMENT SERVICE 5: provision of decrypted information.

This list is probably not complete - other services may be offered which, like these five, also achieve legal access in accordance with the OECD Guidelines and Annex E.

"A trust service" is one in which either:

• the service provider executes a cryptographic operation to the benefit of the subscriber (which would, for example, cover certification and timestamping);
• the service provider is trusted by the subscriber to maintain the confidentiality of an item of information in accordance with a security policy which is published or agreed with the subscriber;
• the service provider is trusted by the subscriber to use a cryptographic variable only in accordance with a security policy which is published or agreed with the subscriber; or
• the service provider offers services to the subscriber which also enable legal access to message decryption keys."

1. This definition, and the services excluded from licensing, would then depend on a further definition of the term "subscriber". A permanent or contract employee, supplier company, and so on, would not be considered a subscriber by virtue of those relationships. (They may be subscribers by virtue of other relationships - for example a licensed service provider may offer discounted trust services as an employee benefit.) A customer would not necessarily be regarded as a subscriber if the main objective of the contract was to provide the customer with goods or services unrelated to trust services. By extension, a consortium of companies built mainly for purposes other than the delivery of trust services would not contain any "subscribers" even though communication internal to the consortium relied on trust functionality. A closed user group, exempt from licensing, can be defined by the absence of "subscribers" within it.
2. The definition is also useful in understanding how a trust infrastructure internal to a company or closed user group should be treated when it cross-certifies with the licensed infrastructure. The consultation paper proposes that this single action would require the whole of the internal infrastructure to become licensed. The analysis argued for here would come to an opposite conclusion.
3. Two things might happen in such a cross-certification:

1. The private key of the internal infrastructure would be used to sign the public certificate of the licensed provider.
1. The licensed provider would certify the root public key of the internal infrastructure.

Action 1 could take place at any time, without the approval, or even the knowledge, of the licensed provider. No licensing regime could sensibly prohibit such an action. When the action is complete, every user of the internal infrastructure would be able to authenticate every user in the licensed infrastructure.

Action 2 is no different in principle from a licensed provider certifying a key that is not a root key - the difference is that the licensed provider has a corporate, rather than an individual, subscriber and that a group of users, instead of a single user, becomes fully part of the licensed infrastructure, authenticatable to every user of the licensed infrastructure.

1. We comment in this section on the benefits that DTI perceives for the proposals, and which are set out in paragraph 42 of the consultation paper thus:

1. This benefit is well stated, but applies mainly to users who, at the inception of the regulatory regime, do not already have existing satisfactory trust relationships.

• TTPs will be able to offer interoperability of secure services hitherto unavailable - While encryption products are available in the UK domestic market, interoperability between different products is not possible. A lack of standards for algorithms and interfacing, coupled with the high burden of key management has created a fragmented market. Today's encryption market is thus very costly. Passing the key management to TTPs and building products to a common architecture will allow access to cryptography for everyone with a PC. Encrypted communications, therefore, will no longer be limited to Governments and larger organisations.

1. The main reason for a fragmented UK market for encryption products has been the attitude of government up to now to the non-government use of cryptography, together with export control policies in countries where products are developed - particularly the United States. Key management presents few "burdens".

 

2. Neither the development of a TSI, however important such a development is, nor mandatory licensing of service providers, will address the issue of standardisation of algorithms and interfacing. Instead, when the general use of key restoration solutions encourages governments to change attitudes and relax export rules, the market will quickly resolve any standards issues.

• TTPs will allow UK Business to take advantage of secure electronic trading- The wide availability of cryptography will allow more paper based transactions to be conducted electronically. Time stamping, non repudiation, confidentiality, authentication and integrity are all necessary to install trust in the electronic age and to allow electronic contracts to take the place of paper ones.

1. It is possible to take some advantage of secure electronic trading without the development of trust services. There can be no doubt, however, that the availability of services makes the advantages significantly greater. The main business beneficiaries will be SME's; larger organisations have options for establishing trust with their business partners and with each other that do not depend on public service provision.
2. A longer example list of trust services might include:

• Secure contract establishment, and user registration;
• Assignment of globally unique distinguished names;
• Generation and allocation of key pairs for key restoration;
• Where requested, generation of key pairs for key exchange;
• Certification of public keys for:

1. digital signature;
1. key exchange; and
1. key restoration;

• Secure storage of key restoration keys and other information;
• Retrieval, when authorised by the contract or by law, of keys and key restoration information;
• Management of lists of revoked certificates;
• Access to, and management of, directories;
• Non-repudiation services;
• Guaranteeing information currency through time stamping; and
• Notarisation of electronic negotiable and legal documents.

1. It is not clear that licensing considerations are the same for all these services.

• TTPs will also be able to offer Data recovery Services - At present, if encryption keys are lost, stolen or deliberately withheld by disaffected employees then the information will remain encrypted and may be lost to its owner for ever. TTPs will be in a position to offer recovery of the keys to their clients as they will store (or escrow) the keys.

1. Key restoration for stored data is without doubt an important service and will certainly be offered to subscribers, but the mechanism is not as simple as the paper assumes.
2. As we have explained, the most convenient way of exchanging confidentiality keys, even between two strangers, often does not need the intervention of a third party; for stored data, where there is only one user and therefore no exchange of keys at all, then it is certainly the case that no other party is required for the simple operation of encryption.

 

3. It is likely that users will choose to be able to invoke key restoration services for stored encrypted data, but will do so by implementing a product- and services-based framework which gives effective and reliable key restoration through mechanisms which complement encryption functions, rather than simply exploiting mechanisms for key exchange. Key restoration services will only be relevant to stored data when the product performing the encryption is specially engineered to prepare key restoration parameters for storage with the encrypted data.

• Encryption policy will help UK manufacturers to export robust encryption within their products - Products that are designed to operate within a TTP environment will be subjected to simpler export licence procedures, thus allowing them to be exported with minimum restrictions. This increased availability will stimulate a greater demand for encryption products both in the UK and foreign markets. UK companies should be in a good position to take advantage of this.

• Use of licensed TTPs is voluntary - those wishing to do otherwise are at liberty to do so - The market will decide if it wants to use TTP services and not Government. The Government believes that the benefits of this scheme will far outweigh any others. Of course those wishing to use any other cryptographic solutions can continue to do so, but they will not be able to benefit from the convenience, and interoperability of licensed TTP services.

1. We expect that there will be a significant and growing market for trust services. We fully endorse the policy that use of trust services should be voluntary.

• UK taking lead in very important area - Many countries are currently trying to develop a cryptographic policy. Many countries agree with the UK that widespread use of cryptography must not be to the detriment of law enforcement requirements. The Government believes that this scheme is the best way to achieve this balance and that other countries may also see the benefits and follow the UK lead.

1. We applaud the lead that the UK is taking and we are confident that other countries will wish to learn from the UK experience.

• Help to increase the Volume of electronic trading - Transactions over the Internet are forecast to reach as much as £22 billion by 2005. Some research suggests that the Internet will take 5 -10% of all retail traffic by 2000. Various barriers (lack of security products and standards, and poor interoperability) may impede this growth. The introduction of TTPs should help in this respect.

With five reservations, we would be willing to accept an initial regulatory regime as proposed. We hope that the Government would wish to keep the arrangements under constant review so that difficulties in the scheme's implementation could be quickly addressed.

Our reservations are:

• The term "encryption service" needs to be carefully defined (and we prefer the term "trust service").
• We would expect industry to be consulted before the Secretary of State fixed or varied general licensing conditions.
• We would argue for a much wider definition of closed user groups exempt from licensing.
• We would like to see emphasis on the attractiveness, acceptability and, hence, take-up of any implemented scheme especially where SMEs are concerned. tow of the key factors that will determine take-up are cost and ease-of-use.
• we would want to understand how any costs of the licensing schemes are to be defrayed.

2. Paragraph 65 - Where views are sought on the proposed conditions.

With three reservations, we believe that the proposed conditions will implement the right level of rigour and consumer protection.

Our reservations are:

• The conditions will not apply to the exemptions we have identified - such conditions would place a heavy burden on a closed user group.
  
• Trust functions should not always be isolated from other business functions. One of the major tasks of a service provider will be managing risk and processing customer registration. Many companies (banks, insurance companies, etc.) will already, through their normal business, understand the risk their customers present and this information will be an essential input into the risk management of a trust service.
  
• We would wish the technical assurance to be adequate but pragmatic. This means that we would be opposed to elaborate procedures dependent on formal evaluation.

The prohibition should relate explicitly to cross-certification, not to making a contract. There should be a legal requirement which mandates every licensed service provider to cross-certify with as many other licensed providers as are necessary to ensure that there is only one public licensed infrastructure in the world. When considering cross-certification with an unlicensed provider, a licensed provider should be required to take into account the integrity of the whole global licensed infrastructure as well as any national or international legal obligations.

May, 1997

Enquiries regarding this response should be addressed to:

C W Binns

FEI

Russell Square House

10 -12 Russell Square

London WC1B 5EE

Tel: 0171 331 2021

Fax 10171331 2042

Email cbinns@fei.org.uk