Intelligence Agencies: Why Do Democracies Build Secret Institutions?

What is an intelligence service?

Intelligence work is often equated with spies, wiretaps or secret operations. Its core is broader: collecting information, analyzing it and delivering it to decision-makers. DCAF describes intelligence services as state organizations whose principal tasks are the collection and analysis of national-security-related information and its dissemination to decision-makers. Some countries separate foreign and domestic services; others place parts of the mission in police or military organizations.

The boundary between intelligence and political authority is equally important. The CIA, for example, publicly states that it collects foreign intelligence and produces analysis but does not make policy and is not a law-enforcement organization. Modern U.S. analytic standards explicitly require analysis to be objective and independent of political considerations. That is a normative standard, not proof that politicization can never occur, but it marks a central democratic distinction: intelligence should tell a decision-maker what the best available evidence suggests, not what the decision-maker is politically supposed to want.

Collection is not only secret spying

Some intelligence is acquired through classic clandestine methods: human sources, intercepted communications, technical sensors, satellite imagery or covert surveillance. But an important part of useful intelligence comes from open sources as well. The U.S. Intelligence Community's OSINT strategy defines open-source intelligence as intelligence derived exclusively from publicly or commercially available information that addresses a specific intelligence priority, requirement or gap.

That corrects the cinematic picture of intelligence. A secret is not automatically the best information. A publicly available satellite image, shipping registry, local newspaper, academic paper, corporate filing or social-media post may reveal more in the right context than a dramatic clandestine source. The system's task is to combine different streams, assess credibility and identify what is still unknown.

Analysis is not prophecy

Intelligence services do not exist because they can eliminate uncertainty. They exist to reduce it and describe it more accurately. Analysts often work with incomplete, contradictory and deliberately deceptive information. They therefore have to distinguish fact, source reporting, assumption and analytic judgment, and express uncertainty clearly. U.S. ICD 203, for example, requires analysts to describe source quality, explain uncertainty, distinguish intelligence information from assumptions and consider alternative explanations.

This also means intelligence is not omniscient. A good service can warn of a possible attack and still miss the timing; it can know an adversary's capability without knowing intention; it can possess many pieces without connecting them soon enough. Calling something 'intelligence' therefore does not make it certain. At its best, intelligence is disciplined judgment under uncertainty.

Counterintelligence: who watches those who are watching?

Another core function is counterintelligence: detecting and limiting the activities of foreign intelligence services, their agents, infiltration, technology theft, covert influence and other attempts to acquire protected information. It is the mirror image of collection. A state that seeks the secrets of others must expect others to seek its own.

Counterintelligence becomes especially sensitive when it touches domestic politics. A democratic system must sharply distinguish genuine foreign intelligence activity, violent threats or covert influence from lawful political opposition. When the label 'security threat' expands to cover opposition groups, journalists or social movements without adequate legal grounds, an intelligence service can begin to resemble a political police force.

Covert action is not the same thing as intelligence collection

A particularly important distinction separates intelligence collection from covert action. In the first, the state primarily wants to learn what is happening. In the second, it seeks to influence political, economic or military conditions while concealing or not publicly acknowledging its role. U.S. law codifies this distinction: covert action is activity intended to influence conditions abroad where the U.S. role is not meant to be apparent or publicly acknowledged, while activities whose primary purpose is intelligence acquisition are excluded from the definition.

That distinction matters because secret analysis and secret intervention are not the same function. Democratic oversight must therefore ask different questions: who authorized the action, what objective it serves, what legal basis exists, who was notified, what risks are involved, and whether political responsibility can ultimately be traced to those who ordered it. In the United States, covert action requires a presidential finding and congressional intelligence committees are generally required to be informed.

Why secrecy is genuinely necessary

If a state publicly disclosed the identity of a human source inside a foreign government, that source might lose access, liberty or life. If it revealed a technical interception method, the adversary would change communications. If it announced in advance whom it was monitoring over an impending attack, the target could disappear or destroy evidence. Operational secrecy therefore has a direct functional purpose: it prevents the observed party from adapting behavior to the observation.

Internal compartmentation also has a rationale. A 'need-to-know' rule can limit damage if an insider is compromised or a document leaks. Yet secrecy has a cost. An overly compartmented system can prevent analysts from seeing an important part of the whole. Effective services therefore have to balance source protection against sufficient sharing to connect dispersed fragments.

Why the same secrecy is democratically dangerous

The structural problem is simple: an ordinary citizen cannot assess an operation they do not know exists. A journalist cannot verify a lawfully classified document. A legislator without authorized access may be unable to see the classified details of a program whose budget is being approved. Without a counterweight, secret powers can expand faster than accountability.

The U.S. Church Committee of 1975–1976 showed that this is not merely theoretical. Its investigation documented abuses involving the CIA, FBI, NSA and other bodies, including domestic surveillance, NSA projects SHAMROCK and MINARET, and COINTELPRO, a program used to disrupt and discredit civil-society and political groups and individuals. Its final findings connected these excesses to the failure to apply constitutional checks and balances, rather than to a single party or president.

That is an important historical lesson: abuse by a secret institution does not require one vast central conspiracy. It can grow incrementally from an overly broad mandate, emergency conditions, political pressure, weak legal boundaries, a culture of impunity or the belief that the end justifies the method.

Oversight has to enter behind the curtain

Because complete publicity is impossible, democracies create indirect forms of scrutiny. DCAF and the Venice Commission emphasize multiple layers: executive control, parliamentary committees, judicial review, independent expert bodies, inspectors, auditing and effective complaint mechanisms. The core idea is straightforward: the public need not know the identity of every source, but someone outside the operational chain must have lawful access to enough information to test legality and proportionality.

Different democracies arrange this differently. In the United States, oversight includes congressional intelligence committees, the Intelligence Community Inspector General and, for specified forms of foreign-intelligence surveillance, the Foreign Intelligence Surveillance Court. In the United Kingdom, Parliament's Intelligence and Security Committee oversees policy, expenditure, administration and operations across the intelligence community; the most intrusive powers use executive authorization plus judicial approval, while the Investigatory Powers Commissioner provides independent scrutiny. Individuals also have access to a specialist tribunal for complaints.

The European Court of Human Rights has repeatedly emphasized a similar principle in secret-surveillance cases: because a person often cannot know during the operation that surveillance is occurring, the law, authorization process, independent supervision, duration limits, data-handling rules and avenues of redress must themselves supply effective safeguards against arbitrariness.

Too little secrecy can fail. Too much secrecy can fail too.

An intelligence system can fail because secrets leak, but it can also fail because it keeps them too effectively from its own people. After the September 11 attacks, the 9/11 Commission found that U.S. security institutions were fragmented, that information did not circulate adequately among the CIA, FBI and other parts of government, and that a culture of 'need to know' often blocked a common picture. The Commission therefore argued for movement toward a 'need to share' approach.

That cannot be absolute either. With no access controls, large stores of sensitive data become easier targets for insider abuse, foreign intelligence services or mass disclosure. The design problem is permanent: share enough to understand threats while limiting access enough to protect sources, privacy and legitimate individual interests.

A secret institution in a democracy is delegated trust

The healthiest democratic formula is neither 'no secrets' nor 'trust us because we cannot tell you.' A better model is divided transparency. The mandate, legal powers, categories of permitted intrusion, budgetary accountability and oversight structure should be public enough for society to understand what authority exists. Operational details, source identities and sensitive methods may remain protected so long as independent people and institutions are legally empowered to examine them.

This is where a democratic intelligence service differs from a secret political police. It should not protect the government of the day from political opponents; it should protect the constitutional order and the population from defined security threats. Its secrecy is a functional exception inside public law, not a hidden source of autonomous power.

The paradox therefore remains, but it need not be mystified. Democracies build secret institutions because some threats operate in secret. Their legitimacy does not come from secrecy itself. It emerges only when secrecy is constrained by law, professional doubt, political neutrality, independent oversight and a real possibility that abuse can be discovered and corrected.