Sir David’s thesis was that it is possible for governments in the liberal democracies to command public support for digital intelligence activities by ensuring that these are conducted in accordance with the rule of law, are properly regulated and entail restraint in the use of State power.
He recalled that under a 2016 Act of Parliament the UK Government can legally authorise, under the necessary conditions, the gathering of intelligence by digital equipment interference, hacking or network exploitation. So it is publicly acknowledged that the UK deploys methods that inevitably carry such moral risks as intrusion into the private lives of the citizen.
It arises from this, he suggests, that it would be desirable to complement the 2016 Act by an ethical code for the secret intelligence profession. Like many other professional codes, an intelligence code can combine consequentialist reasoning (judging the rightness of an act by its results) with deontological prohibitions and personal value ethics involving ideas of honour, integrity and honesty.
To elaborate such a code he collaborated with his colleague, Professor Mark Phythian of the School of History, Politics & International Relations at Leicester, and co-authored a book that Oxford University and Georgetown University Presses have recently published: Principled Spying: the Ethics of Secret Intelligence.
They advocated drawing on Just War doctrine and singled out the following concepts:
- right intention—acting with integrity and having no hidden political or other agendas behind the authorization of intelligence activity or the analysis, assessment, and presentation of intelligence judgments to decision-makers
- proportionality—keeping the ethical risks of operations in line with the harm that the operations are intended to prevent
- right authority—establishing the level appropriate to the ethical risks that may be run and that will then allow for accountability for decisions and oversight of the process
- reasonable prospect of success—having adequate justification for individual operations based on sound probabilistic reasoning that also prevents general fishing expeditions or mass surveillance
- discrimination—determining the human and technical ability to assess and manage the risk of collateral harm, including privacy intrusion into the lives of those who are not the intended targets of intelligence gathering
- necessity—finding no other reasonable way to achieve the authorized mission at lesser ethical risk
The Whole Text of Sir David’s Speech follows
I was delighted when I received the invitation to come back to talk to former Harkness Fellows having done so in 2015. So I can’t have been that boring when I was talking about cyberspace.
Tonight it is a related topic I want to discuss. That of the ethics of digital intelligence. It is a subject I have worked on with Professor Mark Phythian of the School of History, Politics & International Relations at Leicester and the most recent product of our collaboration is our co-authored book that Oxford University and Georgetown University Presses have just published: Principled Spying: the Ethics of Secret Intelligence.
It consists of an extended Socratic dialogue between Mark and myself arguing from our different perspectives about whether ethical secret intelligence is an oxymoron; Mark as a leading political scientist, and myself as a former career practicioner from the worlds of British intelligence, security and defence, coming late to the world of academia. We bring very different experiences to bear.
We chose the dialogue form so that readers interested in the deeply controversial topic of intelligence ethics would be able to follow us as we work our way step by step through each others’ arguments, challenging with counter-arguments and counter-counter arguments to reach our conclusions.
The final chapter of the book is one we wrote together and sets out a set of ethical principles to guide secret intelligence activity, derived from the Just War tradition. It expresses our view that it is possible for governments in the liberal democracies to continue to have effective intelligence support for their decision making – derived from intelligence activities that command public support as being in accordance with the rule of law, properly regulated and with the exercise of restraint in the use of State power for example in limiting invasions of privacy into the private lives of the citizen. Not all democratic states are yet at that point, of course, and the autocracies will never be there, but the UK agencies are I am glad to say leading the way.
I recall a conversation with the head of one prominent Asian-Pacific intelligence agency who remarked to me that he could not understand why the British made so much fuss about such matters. The point of having a secret agency, he said, was that it was secret and did not therefore have to obey the law. It could do whatever was necessary to secure the state from its enemies, free from any public accountability.
Only a few years ago anyone writing about the ethics of collecting and using secret intelligence would have focused just on human intelligence or humint, the oldest part of the profession, going back to Biblical times and before. The purpose of intelligence today is still what it was then: to improve decision taking by reducing ignorance. And the purpose of secret intelligence is to achieve that in respect of information that someone else does not want you to have, and may go to very violent means to stop you having.
[Just think what the Russian GRU was prepared to do in using a sophisticated chemical weapon to try to murder the Skripals in Salisbury. The exotic nature of the weapon, as with the poisoning in the UK of Alexander Livenenko with Polonium 210, a very visible reminder to members of the Russian intelligence community of the fatal consequences of talking to the British.]
If the will of the person with the secret is to be overcome, some form of manipulation or deception or theft or invasion of privacy will be needed. Such behaviours in one way or another contravene everyday ethical norms, and certainly the most uncompromising ethical standard set by the eighteenth-century German philosopher Immanuel Kant via his concept of the categorical imperative. This cautioned that in our lives we should always treat other people as ends and never as means.
To apply this standard to the conduct of intelligence is to set a test it cannot pass. But we conclude that in respect of human intelligence ethical standards are necessary: undercover police officers should not sleep with and father children by members of the protest group they are investigating. Agents should not be placed at unnecessary risk, their interests should not be betrayed and they and their families must be looked after even long after their service to us is over. Blackmail incidentally is incidentally not a good tactic for recruiting: the resentful victim will take delight in misleading and spend all their time working out how to get out of the arm-lock. British intelligence officers know that a reputation for behaving well, even when encouraging betrayal by others, is essential. In that respect the world of John Le Carre is far from today’s reality
But I am getting ahead of myself.
I suggested that thinking about intelligence ethics a few years ago would have revolved around human agents: Today, digital intelligence is at the heart of modern intelligence and it too poses considerable ethical dilemmas. Let me explain why I say that.
Since the days of the Cold War many new demands for intelligence have arisen, for example on the subversive agendas of dictators and autocrats, on jihadist and other terrorists, on cyber criminals and other serious international criminal gangs, on proliferators and pirates, money launderers, pedophile networks and on all those who mean us significant harm.
Many of those demands are about intelligence on people, their identities, location, movements, networks, financing, capabilities and intentions. And where do we look today for information about people? The sources of supply are mostly digital. All that information on the Internet that we unknowingly give to social media. All those web-sites and digital data-bases that can be mined for information. Such as airline data bases and advance passenger records, immigration and passports, credit cards and financial transactions, fingerprints and DNA etc etc. And all the global communications data about who contacted whom, when, where, and how. Intelligence officers today need to understand how the digital world actually works, with mobile devices, social media, bulk data, and ever more sophisticated means of interrogating data with artificial intelligence, machine learning, Bayesian causal networks and search algorithms and so on. The Internet of Things – such as your Internet connected motor car, or toaster, fridge or child’s doll, will all be generating a vast supply of data about their users and their homes.
If the demands for intelligence are always changing so are the digital sources of supply, even more rapidly.
That dynamic interaction between changing demands for intelligence and changing sources of digital supply for intelligence needs to be modulated by a third force, that of the constraints of law and ethics.
When I graduated from Cambridge in 1969 I joined GCHQ in Cheltenham as a young fast streamer. I joined through a shadowy recruitment process, they used a cover story about what they did, a fiction which I was supposed to use with my parents and friends – the fact that it was the UK’s largest secret intelligence agency, continuing Bletchley Park’s role of intercepting and decrypting global communications was itself a deep secret, not avowed to Parliament or public, and a similar wizard’s invisibility cloak was spread over MI5 and MI6. As Professor Christopher Andrew of my old College, Corpus Christi has described it, one result was serious historical attention deficit disorder. There was a missing dimension of history.
Compare and contrast with today. An official history of GCHQ for the 100th anniversary has been commissioned from Canadian Professor John Ferris of Calgary University. The founding documents of the 5-eyes (US/UK/Australia/Canada/New Zealand sigint partnership are openly on the web. The GCHQ website invites applications for new recruits. GCHQ has an iconic donut building. My successors as Director give public speeches. GCHQ sponsors mathematical and technological work in the universities and runs an accelerator programme for private-sector cyber-security start-ups.
GCHQ has always had a special role in keeping UK military, diplomatic and financial communications secure through its expertise in cryptography. Very significantly, the UK now has a new National Cyber Security Centre, based openly in a futuristic office building opposite Victoria Station in London, and announcing itself as proudly part of GCHQ. If you can collect digital intelligence on the nation’s enemies you will be best place to advise on how stop them collecting on all of us.
All that represents massive cultural upheaval in my working lifetime. And the pace of change continues as world events and digital technology in all its forms continue to surprise us. It also means that for the UK the boundary of genuine secrecy has been redefined. For example, under the new 2016 Act of Parliament the UK Government can legally authorise, under the necessary conditions, the gathering of intelligence by digital equipment interference, hacking or network exploitation, all activities previously covered by ‘NCND: ‘neither confirm nor deny’.
So the real secret now is which of those intrusive methods of gathering intelligence are being applied to which target, not over the general existence of the method itself. So it is publicly acknowledged that the UK deploys methods that inevitably carry moral risk such as intrusion into the privacy of those not the legitimate target of an investigation. Intelligence therefore represents a distinctive ethical realm.
All professions have codes of ethics. Having one is close to a defining condition for a profession to be accepted as such. Professional codes of ethics are often incorporated into enforceable regulation. As a result, we are more inclined to trust solicitors not to embezzle, teachers not to seduce students and scientists not to cheat on their results. What, then, are the ethical standards that we want the intelligence profession to uphold and the ethical considerations we want intelligence officers to operate by as they work on our behalf to protect society from those who wish to cause us harm? What it is right for intelligence officers to do in their pursuit of other people’s secrets? How should this activity be regulated and overseen?
These questions sit at the heart of broader debates concerning the relationship between the citizen and the state. Publics need, want—and demand – protection from a range of serious threats, including those posed by terrorism and serious criminality. They also want to be safe in using cyberspace and to have active foreign and aid policies to help resolve outstanding international problems. Secret intelligence is widely accepted as essential to these tasks and a legitimate function of the nation-state.
It also helps in thinking about these modern digital dilemmas to be reminded of the classic traditions of moral philosophy that provide the frameworks by which we have traditionally consider ethical questions and navigate ethical dilemmas. Today’s professional ethical codes are mostly a mix of teleological consequentialist reasoning (judging the rightness of an act by its results such as doctors saving life); deontological prohibitions and injunctions imported from religious or other codes (such as the commandment, thou shalt not commit murder); and aretaic personal value ethics (how one decent human being should behave towards another, involving ideas of honour, integrity and honesty).
The historical tradition for intelligence agencies has been consequentialist, justifying to themselves and their political masters the ethical rightness of their actions by the harms that their interventions prevented. That leads naturally to discussion of the idea of proportionality – that the ethical risks run by an operation should bear a relationship to the extent of harm to be prevented.
This gives rise to the principle that the greater a danger to society the greater the ethical risks that can to be justified to prevent it. This also leads to discussion of the part played by understandings of imminence in reaching these judgements. Crudely put, an armed police officer can open fire if the threat to the public is imminent – but not shoot a wanted terrorist suspect on sight if no threat is being posed.
This, in turn, highlights the ethical problems that arise from the preventive paradigm in which intelligence operates; it is concerned with finding out about and preventing things that have not yet happened (and so may never happen – a possibility that increases, the less imminent the risk).
In balancing these complex judgements, I have found in my classes at King’s College and at Sciences Po in Paris where I also teach that almost all students on reflection resile from the proposition that the prevention of very serious harm to the public, for example from a ‘ticking bomb’ about to detonate in a crowded city, would provide moral justification for maintaining state torturers able to be called in to extract from the suspect the bomb’s location. So the students can see the case for the importation of the human rights canon into the consequentionalist tradition. In that way human rights can be used to provide an essential deontological end point to any consequentialist slide into ever more ethically risky behaviour.
A proposition of John Stuart Mill’s that is relevant to thinking about the ethics of intelligence and I find provokes lively classroom discussion is that: “A person may cause evil to others not only by his actions but by his inaction, and in either case he is justly accountable to them for the injury.” Thus, for example, holding back on using intrusive methods to collect intelligence on a terrorist suspect out of respect for their privacy carries ethical consequences, and those who imposed the restriction would be answerable should the suspect go on to commit an act of terrorism. This is the territory of front page media allegations of intelligence failure after a terrorist attack when it is revealed that the perpetrator was previously known to the authorities. In my book, the real intelligence failure is if an attack is conducted by a terrorist that had escaped all knowledge of the authorities.
A key question in exploring the relationship between intelligence, the state, and the citizen is how far we can speak in terms of an inherent tension between law and secret intelligence? The rule of law is a fundamental principle by which liberal democracies set store, yet intelligence activity with its moral exceptionalism does not easily fit under the rule of law. Conversely, just because a valid legal justification can be found for an intelligence operation does not necessarily make it ethically sound (or politically wise).
Useful insights that can be applied to the practice of intelligence can be drawn from studying how over the centuries the Just War philosophers have sought to reconcile the essential wrongness of causing harm to others with the essential rightness of protecting their community and the innocent and weak from the aggression of others.
The ideas of the justice of war (jus ad bellum) and justice in war (jus in bello) can be seen as having intelligence equivalents (jus ad intelligentiam and jus in intelligentia). We see these as concepts to be used intelligently to unpack the ethical dilemmas, not as a check-list to be ticked off on the way to approving an operation.
In brief summary, in our book we set out the following Just War derived jus in bello concepts to help thinking and teaching about intelligence ethics:
- right intention—acting with integrity and having no hidden political or other agendas behind the authorization of intelligence activity or the analysis, assessment, and presentation of intelligence judgments to decision-makers
- proportionality—keeping the ethical risks of operations in line with the harm that the operations are intended to prevent
- right authority—establishing the level appropriate to the ethical risks that may be run and that will then allow for accountability for decisions and oversight of the process
- reasonable prospect of success—having adequate justification for individual operations based on sound probabilistic reasoning that also prevents general fishing expeditions or mass surveillance
- discrimination—determining the human and technical ability to assess and manage the risk of collateral harm, including privacy intrusion into the lives of those who are not the intended targets of intelligence gathering
- necessity—finding no other reasonable way to achieve the authorized mission at lesser ethical risk
Privacy for example is not an absolute right comparable to the right not to be subject to torture. The validity of a balancing act against other rights, such as the right to security, will therefore depend crucially on the filtering and selection algorithms being sufficiently targeted, using seeds or precise search criteria, to have a reasonable prospect of success in providing important intelligence leads that would not otherwise have been obtainable (thus also passing both the tests of proportionality and necessity). As digital intelligence methods continue to develop (including the application of advanced facial recognitions systems and data from the Internet of Things) such reasoning will become even more important in order to avoid unacceptable violation of the privacy of the innocent and especially a slide into ‘mass surveillance’ which would be unlawful in the UK as well as unethical.
The reassurance that intelligence agencies and professionals within them are behaving in manners that the public would consider acceptable is provided, in the context of secrecy, by the operation of oversight and accountability mechanisms.
The requirement to exercise restraint in the use of the coercive powers of the state is an ethical injunction: not every intelligence operation that may be possible and can be made lawful should necessarily be carried out. Each requires consideration of the potential gain set against the ethical risks being run (for example to the collateral invasion of personal privacy of those not the target of the operation). It is an important part of their recruitment and subsequent education that intelligence officers are helped to develop ways of behaving well under difficult circumstances, even when knowingly encouraging betrayal or intruding on the privacy of private communications and family life.
Experimental psychology and experience in war nevertheless both demonstrate that even those who see themselves as highly moral actors can be led to behave in unacceptable ways when placed in an unhealthy environment. So ethical issues in intelligence have a situational as well as a personal dimension, not least when it comes to designing organizational structures, and statutory democratic safeguards and internal processes to ensure that future governments cannot misuse the powerful intelligence capabilities that intelligence and security agencies must continue to possess.
I like to think that we are now evolving a social contract between the public, their representatives in elected government and the intelligence agencies, and the private corporations that increasingly manage and use our personal information. In the end, liberal democracies need the consent of the public to confirm the ethical soundness of the activities carried out in our name and in their interest. I support therefore a 3-R approach: All activity should be conducted within the rule of law, there is regulation and proper democratic accountability through both judicial and legislative oversight, and authorities should exercise restraint to respect the privacy of the individual and apply the principles of proportionality and necessity at every stage.
Thank you again for the opportunity to discuss these issues with you