An overview of the early development of the UKUSA Agreement, a broad framework which emerged from a series of postwar signals intelligence sharing agreements to form the basis of the Five Eyes alliance.
Forged during the Second World War and its immediate aftermath, the intelligence sharing agreements between members of the Five Eyes alliance (US, UK, Canada, Australia and New Zealand) have been unparalleled in scope and ambition. These agreements, initially referred to as the BRUSA Agreement and later consolidated under the UKUSA Agreement, have provided the basis for an enduring and expansive intelligence sharing partnership between the five countries which has lasted almost 80 years.
Declassifications by Five Eyes intelligence agencies since 2010, as well as documents disclosed through Freedom of Information challenges, provide an important window into the development of this alliance in the immediate postwar years. The formalisation of this postwar partnership began in 1946 as a bilaterial alliance between the UK and the US, committing both parties to a near ‘unrestricted’ exchange of signals intelligence (SIGINT). Canada, Australia, and New Zealand initially held subordinate positions within a system of ‘Empire Intelligence’, which was at this time firmly directed from London. But over the next ten years, the three former Dominions gained varying degrees of ‘SIGNIT sovereignty’, gradually integrating into the partnership on more equal terms. As communications intelligence (COMINT) evolved during the early Cold War, intelligence sharing between the Five Eyes countries quickly outpaced the technical terms of the UKUSA Agreement, which soon evolved into a broad framework committed to the principle of the unrestricted sharing of intelligence.
Forging the 1946 BRUSA Agreement
The 1946 BRUSA Agreement was a landmark bilateral intelligence sharing partnership signed by representatives from the UK’s London Signal Intelligence Board (LSIB) and the US State-Army-Navy Communications Intelligence Board (STANCIB). Both organisations specialised in the interception and decryption of signals and communications intelligence, and the Agreement emerged from the experience of US-UK intelligence cooperation during the Second World War. In this context, the Holden Agreement had, in October 1942, established intelligence sharing between the US Navy and the UK’s Government Code and Cypher School (soon to be renamed as the Government Communications Headquarters, GCHQ), and in May 1943 an initial BRUSA Agreement had extended the scope of Anglo-American intelligence sharing in relation to the war against the Axis powers by regulating procedures for the exchange, security, and dissemination of intercepted information. Following the end of the Second World War it was felt amongst some that a new agreement was needed for peacetime intelligence sharing. As a result, US President Harry Truman signed a memo in September 1945 authorising US military chiefs to continue collaboration with their British counterparts in the field of communications intelligence.
US-UK intelligence sharing during WW2
These declassified files contain a number of documents relating to UK-US intelligence sharing during the Second World War. These include:
- A bundle of US Government papers dated 1940-1944 concerning requests, discussions and agreements to exchange secret technical information with the British.
- A July 1944 memo setting out proposed ‘collaboration of US and British radio intelligence organisations on Japanese and German projects.’
- The text of an agreement reached in May 1943 between the Government Code and Cipher School (the predecessor to GCHQ) and the US War Department, which covered ‘the production, exchange and dissemination of all special intelligence derived by cryptanalysis of the communications of the military and air forces of the Axis powers, including their secret service,’ and provided for ‘the complete interchange of technical data and special intelligence from the sources covered.’ This agreement was, it was noted, ‘extremely broad and apparently is intended to result in complete collaboration,’ including the ‘complete exchange of all intelligence obtained.’
- A series of memos from 1944 relating to the newly-established ‘BRUSA Circuit’, connecting stations in Washington, Pearl Harbour, Melbourne, Colombo and London. The circuit was designed to ‘carry all Technical Information, all Decryption Intelligence and all Traffic Intelligence considered worthy of dispatch by any of the five stations concerned.’
Discussions on a more permanent agreement were held in Washington in October and November 1945, with minutes showing that British intelligence officers were strongly in favour of continued close collaboration, working hard to soothe the suspicions of their American counterparts, particularly in the US Navy. In the first meeting between the two parties, Sir Edward Travis, Deputy Director of the Government Code and Cypher School, noted that ‘the most desirable type of collaboration to be achieved [was] a partnership… [since] the field of communication intelligence is not readily adaptable to the separation of its several branches and that any cooperative effort will be severely weakened by any limitations to full collaboration.’ The minutes show that Travis ‘recommended that complete partnership with mutual access to all branches of communication intelligence and on all tasks be accepted as a basic principle of cooperation,’ though he recognised that ‘there might be specific tasks regarded by either party as purely “domestic” problems and that such tasks might wisely be reserved as exceptions to the partnership.’ Even these exceptions should be agreed upon, however, with Travis stating that ‘if there were to be any reservations, they would be “open reservations” subject to the knowledge and agreement of both parties.’ When inter-agency division within STANCIB looked to challenge these principles of cooperation, Travis again made the British position clear, noting in a joint meeting in November 1945 that ‘he had come to Washington with authority from the London Sigint Board to arrange complete (100%) collaboration… and that any exception thereto can only lead to suspicion… [and] will reduce the working efficiency of all parties.’
The signing of a new BRUSA Agreement in March 1946 thus represented an important moment in the development of the Anglo-American ‘special relationship’, forming the underlying basis for intelligence cooperation throughout the Cold War. The Agreement specifically related to the sharing of intelligence gleaned from the interception of ‘foreign communications’, which were defined as ‘all communications of the government or of any military, air, or naval force, faction, party, department, agency, or bureau of a foreign country, or of any person or persons acting or purporting to act therefor, and shall include communications of a foreign country which may contain information of military, political or economic value.’ The potential sources, therefore, were vast, but they were not without limits. While the Snowdon revelations in 2013 raised concerns about domestic surveillance by members of the Five Eyes alliance, the 1946 BRUSA Agreement was explicit in stating that the US, the British Commonwealth of Nations, and the British Empire were excluded from this type of intelligence collection and sharing.
In depth: the 1946 British-US Communications Intelligence Agreement
The March 1946 British-US Communication Intelligence Agreement, as avowed and declassified for the first time in 2010, is a central document in the history of US-UK (and later Five Eyes) intelligence sharing.
Classified as Top Secret, the agreement applied to all US and British intelligence agencies, including all those throughout the British Empire (excluding the Dominions). The parties to the agreement agreed to share all intelligence gathered relating to all ‘foreign countries’, taken to include ‘any country, whether or not its government is recognised by the US or the British Empire, excluding only the US, the British Commonwealth of Nations, and the British Empire.’ Crucially, such wide-ranging intelligence exchange would be ‘unrestricted on all work undertaken except when specifically excluded from the agreement at the request of either party and with the agreement of the other. It is the intention of each part to limit such exceptions to the absolute minimum and to exercise no restrictions other than those reported and mutually agreed upon.’
As well as such unrestricted sharing between the two countries, each agreed to keep the existence of the agreement secret from all ‘third parties’, and to only enter into additional sharing agreements with these third parties with the express permission of the other. Any intelligence shared with third parties would, in turn, be shared in secret with the other side.
The BRUSA Agreement was accompanied by a set of appendices, which developed following a joint technical conference in March 1946, providing important detail on the forms and means of collaboration across the field of signals intelligence:
- Appendix A: Security and Dissemination Regulations
- Appendix B: Coordination and Exchange of Information on Cryptanalysis, Traffic Analysis, Decryption, Translation and Associated Techniques
- Appendix C: Collection and Exchange of Raw Traffic
- Appendix D: Exchange of Communication Intelligence
- Appendix E: Exchange of Collateral Material
- Appendix F: Communications
- Appendix G: Liaison and Channels for Exchange
A later technical conference, in July 1948, resulted in amendments to these appendices, although the text of the overarching agreement remained unchanged.
The Agreement also covered a number of other related areas, including: the acquisition of communications documents and equipment; cryptoanalysis; decryption and translations; and the acquisition of information regarding communication organisations, practices, procedures, and equipment. Importantly, the exchange of information was understood to be ‘unrestricted’. While information considered to be in the national interest could be ‘specifically excluded’ at the request of either party and with the agreement of the other, the Agreement maintained that ‘it is the intention of each party to limit such exceptions to the absolute minimum and to exercise no restrictions other than those reported and mutually agreed upon.’
Security and secrecy were afforded the highest priority in negotiating the terms of the 1946 BRUSA Agreement, with Appendix A and Appendix B defining the rules governing the dissemination of intelligence. Information shared as part of the Agreement was restricted to codeword-classified documents, accessible only to those who had been ‘indoctrinated’ into the agreed security procedures. Nevertheless, the ambitious nature of the Agreement was clearly highlighted in Appendix C, which noted that: ‘In general the procurement, maintenance, and utilization of rapid means of exchange and of rapid means of delivery from intercept points shall be promoted and encouraged by each party to the utmost extent of its means, and usable radio teletype channels shall be kept filled to the limit of their capacity with raw traffic.’
In the years immediately following the 1946 BRUSA Agreement, the terms of intelligence sharing between the US and the UK evolved through a series of technical conferences – in March 1946, July 1948, November 1950, and March 1953 – each producing a revised set of appendices to the initial Agreement. While the 1946 Agreement provided a framework for unrestricted intelligence sharing, in practice matters were a little more complex. As intelligence historians have shown, a free flow of intelligence certainly existed between the two allies in relation to perceptions of a joint threat from the Soviet Union, but the sharing of material from some other regions – such as Palestine and the emerging state of Israel during the late 1940s – often remained selective on account of competing priorities between the two states.
Canada, Australia and New Zealand: the making of the Five Eyes alliance
While the 1946 BRUSA Agreement concerned bilateral relations between the UK and the US, minutes from the British Signals Intelligence Conference in March 1946 show representatives from the Dominions of Canada, Australia and New Zealand had been informed of how the Agreement would affect them during a Commonwealth Conference in London that same month. Intelligence officers in London had initially been keen to keep their Commonwealth associates in a subordinate position after the war, with each retaining relatively small peacetime SIGINT units but continuing to conduct their activities under the leadership of the UK as part of a system of ‘Empire Intelligence’.
Indeed, this had been the recommendation of Brigadier General B. Combes, a British army officer on loan to the Australian Department of Defence, who in November 1945 published a report on the organisation of post-war joint intelligence in Australia. Combes’ recommendations were taken up by the Australian Defence Committee, which in July 1946 received approval from the Australian Cabinet for the establishment of a Joint Intelligence Bureau, with New Zealand invited to participate in its activities, and a Signal Intelligence Centre, which would form part of the British Commonwealth signals intelligence effort. Given Australia’s comparatively small SIGINT resources, it was agreed that fitting into the overall Empire intelligence pattern would ‘facilitate the free interchange of intelligence and thus ensure for Australia a world wide coverage of intelligence.’
While Australia and New Zealand were persuaded to continue with this approach, support in Canada for signals independence had been growing following the expansion of Canadian capabilities during the Second World War. Keen to assert their post-war relevance, Canadian intelligence officials were looking for SIGINT sovereignty and their right to make bilateral agreements on their own terms. In negotiating the terms of the 1946 BRUSA Agreement, American military officials had been supportive of independent Canadian capabilities and direct bilateral intelligence-sharing between Washington and Ottawa, since Canada’s geographical position made it an important asset within the intelligence sharing network. Indeed, Article 6 of the 1946 Agreement had subtly recognised Canada’s special status, as while arrangements made by the US with the other Dominions required UK approval, arrangements between Canada and the US only required the US to obtain ‘the views’ of UK authorities.
Following the 1946 BRUSA Agreement, US and Canadian officials began their negotiations for a bilateral agreement. Despite their smaller and comparatively underdeveloped intelligence capacity, Canadian intelligence officials made their demands clear. They wanted access to an official copy of the 1946 BRUSA Agreement, a commitment to an exchange of liaison officers between US and Canadian agencies, and for any ‘need-to-know’ principle included in the CANUSA Agreement to be based on a Canadian assessment of what they needed from the US in terms of SIGINT. They also insisted that any CANUSA Agreement would need to take place through an official exchange of letters between intelligence agencies rather than through a higher level of political approval, most likely because Canadian intelligence officials wanted to maintain control of the process rather than give over to potentially more reluctant political decision-makers.
With these high demands, not all US military officials were onboard. As minutes quoted in a 2019 article by intelligence historian Wesley Wark show, General Agee of the US Air Force made the argument for a more restrictive intelligence sharing agreement. Agee is recorded as noting bluntly that: ‘The Canadians have no information to exchange. The advantages to be gained by complete exchange, are far outweighed by the disadvantage of wider dissemination of information.’ Despite this initial reluctance, American security and intelligence agencies came round to the Canadian proposal. An exchange of letters in 1949 between the heads of the US Communications Intelligence Board and the Canadian Communication Research Committee thus gave independent status to Canada and extended BRUSA signals intelligence sharing into a tripartite agreement.
While the US was establishing bilateral relations with Canada, SIGINT from Australia and New Zealand continued to run through London. A new Signals Intelligence Centre was set up in Melbourne in April 1947 under the guise of the Defence Signals Bureau (DSB). The centre, which operated as a joint UK-Australia-New Zealand organisation run by an integrated staff, was British in form, with a significant number of its officers seconded from GCHQ, including its first Director. The US had initially refused to integrate Australia into the UKUSA framework, as communist infiltration of the Australian Labour government had encouraged a reputation for insecurity that was hard to shake. But the election of the Liberal Party persuaded the US that Australia could be trusted, and the NSA began collaborating more directly with the DSB. A Tripartite Conference was held in Melbourne in September 1953 between UK, US and Australian officials, the main recommendations of which were incorporated into an amended Appendix J (Principles of UKUSA Collaboration with Commonwealth Countries other than the United Kingdom) and a new Annexure J1 (UKUSA Arrangements Affecting Australia and New Zealand), the latter noting that the DSB would ‘collaborate directly’ with the NSA.
Prior to this conference there had been a surprising lack of clarity among senior intelligence officers about what parts of the BRUSA Agreement and its many appendices had been shared with Australia and New Zealand. For example, on 17 June 1953, John Simmonds, Secretary of the London Signal Intelligence Board, sent a letter to John Rendle, Senior UK Officer at the Australian Defence Signals Branch, asking him to ‘do a bit of discreet ferreting for me’ in order to establish what parts of the Agreement were held by the DSB. Rendle’s response, on 9 July 1953, noted that ‘I am extremely hazy about what has happened to the Agreement since it was drawn, but the document we hold is clearly obsolete at most points!’
By the mid-1950s, intelligence officials on both sides of the Atlantic felt that the passage of time had meant that much of the detail contained in the original 1946 Agreement was now outdated. A new version of the Agreement was thus developed and renamed the UKUSA Agreement, a draft of which shows a consolidated version of the gradually updated appendices. However, a minute from the Australian Joint Intelligence Committee from 8 November 1955 suggests that the original 1946 BRUSA Agreement had not been shared with Australian intelligence agencies, though it was noted that Commonwealth countries had nonetheless been working within the technical framework of the Agreement. John Burrough, Assistant Director of the Defence Security Bureau, raised this issue with the Director of GCHQ in a letter on 13 November 1956, in which he noted that: ‘For some time now DSB has been uneasy about the completeness and current validity of its files relating to the UKUSA Agreement and its Appendices. While appreciating that our Need-to-Know does not extend to all the UKUSA documents, we suspect that there may be gaps in our records of those which we sometimes need to consult.’ Interestingly, the response from GCHQ on 28 February 1957 acknowledged that ‘in order to reply, we had to undertake a thorough overhaul of the Directorate copies of the UKUSA Appendices’, but this had revealed that ‘some of the technical Annexures are still out of line with current practice and there would seem little merit in sending you one set of partially out of date documents to replace another.’ These records show how intelligence sharing between the NSA and GCHQ quickly outpaced the terms specified in the Appendices to the Agreement, which had by now become an integrated working relationship between the two allies.
Having been written into the UKUSA Agreement as one of the ‘collaborating Commonwealth countries’, Australia soon looked to gain greater access to intelligence reports, particularly from the US. On 30 July 1957, the Acting Director of the Defence Signals Branch sent a letter to the Director of GCHQ requesting a wider service of COMINT from the UK and US than had previously been available. Although the letter is heavily redacted, it notes that: ‘While the material already received from GCHQ has been invaluable to Australian consumers and is greatly appreciated, it is observed that it is limited to that originated by GCHQ itself. It seems likely that a more complete service could be given if material originated by NSA could also be released to DSB.’
Given the time involved in keeping the various aspects of the UKUSA Agreement up to date, intelligence officials at GCHQ and the NSA decided after 1961 to abandon this process and instead continued working within a broad interpretation of the Agreement, negotiating issues of a case-by-case basis. As John Ferris has noted in his official history of GCHQ, the agencies ‘ceased updating UKUSA because it had already achieved its aim: a working relationship based on a kind of Comint common law…The UKUSA agreement became the equivalent of a Masonic handshake. Its members relied on trust between themselves, and rules and secrecy in their relations with all outsiders.’
Further reading
Richard J. Aldrich, GCHQ, 2019.
John Ferris, Behind the Enigma: The Authorised History of GCHQ, Britain’s Secret Cyber-Intelligence Agency, 2020.
Wesley Wark, The Road to CANUSA: How Canadian Signals Intelligence Won its Independence and Helped Create the Five Eyes, Intelligence and National Security, 2019.
