Khaled El-Masri was riding a bus toward Skopje on the last day of 2003 when Macedonian border guards pulled him off at the Serbian frontier and told him his passport needed checking. He was a car salesman from Ulm, a German citizen of Lebanese origin, traveling alone on a holiday meant to clear his head after an argument with his wife. The passport check turned into twenty-three days in a locked room at the Skylight Hotel in Skopje, where men in plainclothes questioned him about an Islamic centre and a name, al-Masri, that was close to the name of a genuine suspect but belonged to the wrong man. On the twenty-third day a CIA team flew in, stripped him, beat him, forced a suppository and a diaper on him, chained him to the floor of an aircraft, and flew him to a prison in Afghanistan that its inmates would later learn to call the Salt Pit.

El-Masri was held in Afghanistan for four months. He was beaten, interrogated, force-fed during a hunger strike, and never charged with anything, because there was nothing to charge him with. The Central Intelligence Agency had abducted the wrong person, realized its error within weeks, and then argued internally for months about what to do with a man whose only crime was a surname. Eventually he was flown to Albania, driven up a mountain road at night, and left standing on the verge with his belongings, told to walk and not look back. He thought he was about to be shot. Instead he walked until he reached an Albanian checkpoint, where guards assumed from his appearance and his story that he was a terrorist and nearly arrested him again. El-Masri is the human face of a system, and the system had a name that sounded almost bureaucratic: extraordinary rendition.

Extraordinary rendition was America’s attempt to split the difference between the rule of law and the war on terror. The idea was to seize suspects without arrest warrants, without extradition hearings, without the involvement of any court, and to move them across borders to secret prisons or to the custody of foreign governments that practiced torture, all while maintaining the official fiction that the United States neither tortured anyone nor knew what happened in the rooms where its prisoners were held. The program was meant to be deniable. It was meant to leave no fingerprints. What it actually produced was one of the most thoroughly documented covert operations in modern history, a paper trail of flight logs, hotel receipts, court judgments, and a 525-page summary of a United States Senate investigation that concluded the entire enterprise had been both brutal and ineffective. The story of how the program worked, how it was exposed, and what it cost is essential context for understanding why other states, including India, looked at the American model and chose a different method. The lesson many of them drew was blunt: if you are going to operate outside the law, do not build a system that can be subpoenaed.

Background and Triggers

The word rendition is old, and the practice of moving a suspect from one jurisdiction to another without formal extradition did not begin in 2001. American law enforcement had used a narrow version of it for decades, usually to bring a fugitive back to face trial in a United States courtroom. The 1980s and 1990s produced a handful of these operations against drug traffickers and against men wanted for attacks on Americans abroad. What changed under President Bill Clinton was that the destination stopped being an American courtroom. Beginning in the mid-1990s, the CIA ran what insiders described as a rendition program aimed mainly at members of Egyptian Islamic Jihad and similar groups, seizing them in third countries and delivering them not to the United States but to Egypt, where they faced military tribunals, long sentences, and interrogation methods that no American court would tolerate. Roughly seventy of these transfers took place before the September 2001 attacks. The men were already wanted somewhere. The destination country had an outstanding legal process, however flawed. The pre-2001 model was ugly, but it retained a thin connection to the idea of prosecution.

September 11 severed that connection. Within days of the attacks, the architecture of American counter-terrorism was rebuilt around a single priority, which was the prevention of the next attack rather than the prosecution of the last one. On September 17, 2001, President George W. Bush signed a classified directive, a Memorandum of Notification, that granted the CIA sweeping authority to capture and detain suspected members of al-Qaeda anywhere in the world. The document did not authorize a court process. It authorized custody. The agency now had legal cover, in the view of the administration’s lawyers, to run its own prison system, hold prisoners indefinitely without charge, and decide for itself who was a threat. The pre-2001 question, where do we prosecute this man, was replaced by a new one, what does this man know and how do we extract it.

That shift created an institution that had never existed before, a clandestine detention and interrogation enterprise operated by an intelligence agency rather than by a military or a justice system. The CIA had no prisons, no trained interrogators of the kind the program would require, no detention doctrine, and no experience holding human beings for years at a time. It built all of it from scratch and at speed, and the speed is the key to understanding everything that followed. Decisions that should have taken months were made in days. Facilities that should have been designed by people who understood detention were instead improvised in borrowed buildings. Interrogation methods were adopted not from any tested intelligence tradition but reverse-engineered from a military training program designed to teach American soldiers how to resist torture if they were captured. The program was a series of emergency measures that hardened into permanent practice because nobody in authority was willing to be the official who relaxed the posture and then watched another attack unfold.

The triggers were therefore both legal and psychological. The legal trigger was the September 17 finding and the chain of Justice Department memoranda that followed it, documents that redefined torture so narrowly that almost nothing the agency wanted to do fell inside the definition. The psychological trigger was the certainty, near universal in Washington in late 2001, that a second wave of attacks was imminent and that the only way to stop it was to get inside the heads of the men being captured in Afghanistan and Pakistan. Cofer Black, who ran the CIA’s Counterterrorist Center, told a congressional committee that after September 11 the gloves came off. The phrase captured the mood precisely. A democracy that had spent the Cold War criticizing the Soviet bloc for disappearing people into secret prisons now built its own version, and convinced itself that the difference in motive made the difference in method acceptable.

The legal scaffolding deserves particular attention, because it is what allowed officials to tell themselves the enterprise was lawful. The Justice Department’s Office of Legal Counsel produced a series of memoranda, the most notorious of them signed in August 2002 and associated with the lawyers Jay Bybee and John Yoo, that redefined torture so narrowly as to place almost everything the agency proposed outside the prohibited category. The memoranda argued that physical pain rose to the level of torture only if it reached the intensity associated with organ failure or death, and that mental harm counted only if it was prolonged and lasting. Under that definition, slamming a man against a wall, depriving him of sleep for a week, and pouring water over his covered face to simulate drowning could all be characterized as lawful questioning. The memoranda were later withdrawn and repudiated by the Justice Department itself, and one of their authors would face years of professional controversy over them. While they were in force, however, they served their purpose, which was to convert a set of practices that any plain reading of American and international law would call torture into a set of practices that carried, on paper, an official stamp of legality. The agency had insisted on that paper precisely so that responsibility, if the question ever arose, could be shown to rest with the lawyers who wrote the memoranda rather than with the officers who acted on them.

What made extraordinary rendition extraordinary, as distinct from the ordinary rendition that preceded it, was the removal of the courtroom from the end of the process. The destination was no longer a trial. The destination was a cell, either a CIA-run cell at a secret site or a cell in an allied country chosen specifically because its security services would do things American officers were not supposed to do. The program was designed around deniability. American hands would stay clean because American officers, in theory, would not personally administer the worst of it. This was the central deception of the entire enterprise, and it collapsed the moment investigators understood that arranging torture, transporting a man to it, and writing the questions to be asked during it are not meaningfully separate from torture itself. The architecture India would later study, analyzed in depth in our examination of the global legal debate on targeted killings, began here, in the gap between what the United States did and what it was prepared to admit.

How the Program Chose Its Targets

A covert program is only as disciplined as its target selection, and the rendition program’s target selection was the first place its discipline failed. In the months after September 11 the CIA was receiving an enormous volume of intelligence, much of it raw, unverified, and generated by sources with strong incentives to please. Afghan warlords were paid bounties for handing over suspected al-Qaeda and Taliban members, which created a powerful incentive to label personal enemies and business rivals as terrorists. Pakistani security services swept up hundreds of men in the border regions and presented them to American officers as significant catches. Walk-in informants offered names. Interrogations of early detainees produced more names, including names given under duress by men who would have said anything to make the pain stop. The intelligence pipeline that fed the program was contaminated at the source, and the program had no reliable filter.

The target identification process, as it actually functioned, ran roughly as follows. A name would surface, attached to a claim that the person was a member of al-Qaeda, a facilitator, a financier, or a figure with knowledge of an imminent plot. The Counterterrorist Center would assess the claim, but the assessment was conducted under intense pressure to act rather than to wait, and the institutional bias ran heavily toward capture. A cautious officer who recommended against seizing a suspect, and who turned out to be wrong, would carry the blame for the next attack. An officer who recommended seizing a suspect who turned out to be innocent would carry, at worst, an administrative embarrassment. The asymmetry of consequences guaranteed that doubtful cases would be resolved in favour of capture. El-Masri was a doubtful case. The Macedonian station flagged him, the name resembled a real suspect’s name, and the question of whether to seize him was decided by an officer who, according to the later Senate investigation, had a personal conviction about the case that the evidence did not support.

Once a decision to render a target had been taken, a capture plan was assembled. The plan depended heavily on the cooperation of the country where the suspect was located, because the CIA could not simply operate freely on foreign soil. Cooperation came in several forms. Some governments arrested the suspect themselves and handed him over. Some permitted American officers to take custody on their territory and look away. Some provided intelligence on the suspect’s location and movements and then absented themselves. The program therefore generated a sprawling web of foreign complicity, and that web would later become the mechanism of its exposure, because every cooperating government left its own records, employed its own officials who could later testify, and answered to its own courts and parliaments.

The contamination of the target pool produced a measurable population of innocent prisoners. The exact number will never be known, because the program was designed to leave no comprehensive record and because many of the men swept up were never the subject of any judicial process that would have established their innocence. What is documented is bad enough. The Senate investigation identified multiple cases of detainees held by the CIA who were wrongfully detained, men against whom the agency had no credible basis for suspicion, including individuals held for months after the agency had concluded they were not threats. El-Masri was the most famous because he was a Western citizen who survived to tell his story in European courts. He was not unique. The program’s defenders argued that any system would produce errors. The program’s critics answered that a system without courts produces errors that nobody is obliged to correct, and that this was the precise design flaw, because the courtroom that extraordinary rendition removed was the institution whose job was to catch exactly this kind of mistake.

A bounty system deserves its own scrutiny here, because it injected a commercial incentive into the most consequential decision the whole effort made. In the months after the fall of the Taliban, the United States distributed leaflets across Afghanistan and Pakistan promising substantial cash rewards, in some accounts thousands of dollars per head, for information leading to the capture of al-Qaeda and Taliban figures. In a region where that sum represented a fortune, the leaflets did not merely encourage the surrender of genuine militants. They created a market, and in any market the supply rises to meet the price. Tribal rivals, business competitors, and men with private grudges discovered that an accusation could be monetized. A neighbour who wanted a piece of land, a trader who wanted a competitor removed, a warlord who wanted to please his American paymasters, each had a reason to attach the label of terrorist to a person who had earned it only by being inconvenient. Many of the men who ended up in the detention system, including a substantial share of those later sent to the Guantanamo Bay naval base, entered it through this contaminated channel. An effort that had set out to capture the architects of mass murder had built an intake process that a local feud could exploit.

The target selection failure matters beyond the individual tragedies because it shaped the legal catastrophe that followed. A program that seizes only genuine, high-value terrorists can at least argue necessity. A program that seizes car salesmen on the strength of a similar surname cannot. Every wrongful case became a lawsuit, a parliamentary inquiry, a newspaper investigation, and a piece of the evidentiary record that eventually made the program impossible to defend. The intelligence value of the program, whatever it was, had to be weighed against a target list that included men who knew nothing because they had done nothing.

The Capture: Streets, Airports, and Foreign Cooperation

The seizure itself was the most visible and the most operationally risky phase, and it took place across an astonishing geographic spread. Suspects were taken in the lawless border zones of Pakistan and Afghanistan, where capture often meant little more than accepting custody of a man already held by a local force. They were also taken in the middle of European cities, in the Balkans, in Africa, and in Southeast Asia, in operations that required American officers to function as a kidnapping team in the heart of an allied democracy. The contrast between those two environments defines the program’s range and its recklessness.

The clearest illustration of the European model is the case of Hassan Mustafa Osama Nasr, an Egyptian cleric known as Abu Omar who had been granted political asylum in Italy. On February 17, 2003, Abu Omar was walking to his mosque in Milan when he was bundled into a van on a public street. He was driven to the Aviano air base, a NATO facility in northern Italy, then flown via the American base at Ramstein in Germany to Egypt, where he was held for years and, by his own account, tortured. The Milan operation was a kidnapping conducted by a CIA team on the territory of a treaty ally without the knowledge of that ally’s elected government, although elements of the Italian intelligence service were aware. It was also conducted with a carelessness that proved decisive. The CIA officers involved stayed in expensive hotels, used mobile phones that Italian investigators later tracked, ran up charges on traceable accounts, and left an electronic and financial trail that an Italian prosecutor named Armando Spataro followed with patience and skill. The Milan kidnapping became the first rendition operation to produce criminal convictions, a development examined alongside other accountability cases in our analysis of apartheid South Africa’s death squads and the truth commission that exposed them.

Canadian citizen Maher Arar represents a different capture model, the airport interception. Arar was a telecommunications engineer returning home through New York’s John F. Kennedy airport in September 2002 after a family holiday. American officials detained him on the basis of intelligence, later shown to be unreliable, supplied by Canadian authorities who had linked him loosely to a person of interest. Rather than deport Arar to Canada, his country of citizenship and residence, the United States flew him to Jordan and then drove him to Syria, the country of his birth, where he was held for nearly a year in a cell he described as the size of a grave and subjected to repeated beatings. Arar had no connection to terrorism. A Canadian commission of inquiry, led by Justice Dennis O’Connor, examined his case in exhaustive detail and concluded that he had been the victim of a catastrophic error compounded by torture. The airport model showed that a suspect did not have to be hunted down in a foreign city. He could simply be intercepted at a border crossing he had every legal right to use, and then disappeared.

The third capture model, the handover, was the most common and the least visible. In Pakistan and Afghanistan, hundreds of men passed into American custody not through a dramatic snatch operation but through a transaction. Pakistani forces, acting on their own intelligence or on bounty incentives, detained suspects and delivered them to American officers, often at an air base, often with minimal documentation. The handover model is how figures such as Abu Zubaydah and, later, Khalid Sheikh Mohammed entered the system, although those two were genuine senior figures rather than the contaminated cases that the bounty system also produced. The handover model also explains why the program’s geography was so heavily weighted toward South Asia. Pakistan was both a sanctuary for the men the United States wanted and a state willing, for its own reasons and its own price, to hand some of them over. The dynamics of that sanctuary, and the way it shaped every counter-terrorism campaign aimed at the region, are the subject of our profile of Pakistan’s terror safe-haven network.

The handover model also created a documentation problem that would matter enormously later. A street snatch in Milan, however reckless, at least involved American officers who knew who they had taken and why. A handover at a Pakistani air base often involved a prisoner delivered with little more than a name, a nationality that might itself be a guess, and an unverified claim about his significance. The receiving officers frequently could not independently confirm the most basic facts about the man now in their custody. This is one reason the detention system filled with prisoners whose identities and histories were genuinely unclear, and it is one reason the later effort to determine who in the system was actually dangerous proved so difficult. A capture process built for speed had skipped the unglamorous work of verification, and the result was a population of prisoners about whom the agency itself was uncertain. The uncertainty did not lead to release. It led to indefinite detention, because once a man was inside a secret system that had no court to review his case, no mechanism compelled anyone to decide he should be let go.

Across all three models, the common feature was the absence of legal process. No arrest warrant was served. No extradition request was filed. No magistrate reviewed the evidence. No defence lawyer was permitted to challenge the basis of the seizure. The man was simply taken, and the legal systems of the country where he was taken, the country he was flown through, and the country he was delivered to were all bypassed. This was not an oversight. It was the design. The program existed precisely to operate in the space where law did not reach, and the capture phase was the moment that space was entered. Every subsequent abuse, the transfers, the secret cells, the interrogations, flowed from the foundational decision that the man in question would never see the inside of a courtroom.

The Transfer: The Aircraft and the Black Site Archipelago

A captured suspect had to be moved, and the movement phase is where the program left its most extensive and most damning physical record. The CIA did not maintain its own visible fleet of prisoner-transport aircraft. Instead it relied on a network of shell companies and contractors that operated executive jets, primarily Gulfstreams and Boeing business aircraft, which were registered to corporate entities with anodyne names and innocuous postal addresses. One of the most-studied aircraft was a Gulfstream V that journalists and human rights investigators traced through years of flight logs. These planes flew between American military bases, civilian airports, and the airfields nearest the secret prisons, and every flight generated paperwork, because air traffic control systems, airport handling agents, and fuel suppliers everywhere keep records. The program’s planners appear to have believed that corporate registration and quiet airfields would be enough to hide the operation. They were wrong, and the flight logs became the spine of the journalistic and parliamentary investigations that eventually mapped the entire network.

The transfer itself followed a standardized and deliberately brutal protocol that the agency called, without irony, the rendition process. On handover, the prisoner would be photographed naked, his clothing cut away, a body cavity search conducted, a diaper fitted, and a tracksuit or similar garment put on him. He would be blindfolded or hooded, his ears covered, his hands and feet shackled, and in some cases he would be drugged. He would then be chained to the floor of the aircraft, often in a painful position, for a flight that might last many hours. The protocol was justified internally as a security measure, but it functioned as a psychological one, because it stripped the prisoner of orientation, dignity, and any sense of where he was or what would happen next. By the time he arrived at his destination, the conditioning that the interrogation program depended on had already begun.

The destinations formed what investigators came to call the archipelago of black sites, a scattered chain of secret CIA detention facilities on several continents. The agency assigned the sites colour-based or otherwise neutral cryptonyms in its internal records, and the later Senate report used those cryptonyms in place of the host countries’ names, although journalism, leaked documents, and European court judgments subsequently identified most of the locations. The first major site was in Thailand, where a facility sometimes referred to by the cryptonym associated with the colour green held Abu Zubaydah and Abd al-Rahim al-Nashiri in 2002. A facility in Afghanistan, infamous among detainees as the Salt Pit and identified in official documents under another cryptonym, was the site where a prisoner named Gul Rahman died of hypothermia in November 2002 after being left chained, partly clothed, in a freezing cell. Sites in Poland and Romania, located in former military or intelligence compounds, held high-value detainees in the program’s later years. A site in Lithuania operated for a period. Detainees were also held, at various times, at facilities in or near other cooperating states.

That black site network reveals the program’s central organizational logic, which was the outsourcing of geography. The United States did not want these prisoners on American soil, because American soil meant American courts, the writ of habeas corpus, and the protections of the Constitution. It did not, for the most sensitive prisoners, want them at the Guantanamo Bay naval base either, because Guantanamo had become a magnet for litigation and press attention. The solution was to place the prisons in countries whose governments could be persuaded, through a combination of diplomatic pressure, financial inducement, and intelligence-sharing relationships, to host a secret American facility and to keep its existence off their own public books. Each host country was therefore made complicit, and complicity, once established, is difficult to contain. The Polish, Romanian, and Lithuanian governments would all eventually face investigations, parliamentary inquiries, and rulings from the European Court of Human Rights over their roles, because a covert program that depends on the silence of a dozen sovereign states is a program with a dozen potential points of failure. The contrast with the geographic logic of aerial campaigns waged from the sky, which require no host-country prison at all, is one reason analysts later distinguished the rendition model from the approach traced in our history of the United States drone program in Pakistan.

The economics of the transfer network are worth pausing on, because they reveal how thoroughly the operation had been built to look like ordinary commercial aviation. The aircraft were operated through layers of shell companies, some of them little more than a name, a postal box, and a nominal officer, structures designed so that no casual inquiry would connect the plane to the agency. The planes used regular civilian airports as well as military fields, filed flight plans through normal channels, and paid landing and handling fees like any corporate jet. That was the intended camouflage. It was also the fatal flaw, because ordinary commercial aviation is one of the most comprehensively documented activities on earth. Every flight plan, every landing slot, every fuel purchase, every overflight permission, every crew hotel booking generates a record held by an airport authority, a handling agent, a fuel supplier, or a national aviation regulator, and those records are kept by entities with no stake in the operation’s secrecy. Investigators did not need to penetrate the agency to map the network. They needed only to collect the routine paperwork that the camouflage itself required, and the camouflage, assembled to make the planes invisible, instead left a trail in dozens of countries at once.

Inside the Black Sites

Before a detainee was ever questioned, the black site itself did work on him, and the conditions of detention were a deliberate part of the system rather than an incidental hardship. The cells were small, often windowless, and kept under artificial light around the clock so that a prisoner lost track of day and night. Detainees were frequently held in total isolation, with no contact with other prisoners and limited contact with their guards, for weeks and months at a time. Loud music or noise was sometimes piped into the cells continuously. Prisoners were moved between sites without explanation, hooded and shackled, so that they never knew what country they were in, and many of them, even years later, could not say where they had been held. The disorientation was the point. A man who does not know where he is, what day it is, whether anyone knows he is alive, or whether he will ever leave is a man whose psychological resistance has been worn down before a single question is asked.

The physical conditions at some sites were severe enough to be lethal. The facility in Afghanistan that detainees called the Salt Pit was, by the accounts that later emerged, a place of darkness, cold, and filth. It was there, in November 2002, that Gul Rahman died. Rahman had been left shackled to a wall in a cell, partially clothed, in temperatures low enough to kill, and he died of hypothermia. No one was criminally punished for his death. The officer most closely associated with the decisions around his detention was, according to later reporting, subsequently recommended for a performance award. The Salt Pit case is the clearest illustration of what the absence of oversight produced. A detention facility run in secret, staffed by people who understood that the normal rules had been suspended, with no inspectors, no courts, no journalists, and no families able to ask after the prisoners, is a facility in which a man can freeze to death in a cell and the institution can carry on as though nothing requiring correction has occurred.

The duration of detention compounded the conditions. The program was not built around short questioning followed by transfer to a conventional legal process. Some detainees were held in the black site system for years. Time itself became an instrument, because indefinite detention with no charge, no trial, no sentence, and no end date is a distinct form of pressure, the removal of any horizon a prisoner can orient himself toward. A man serving a known sentence can count down the days. A man held in a secret prison cannot count anything, because no one has told him there is anything to count. This was the environment, isolating, disorienting, indefinite, and at some sites physically dangerous, into which the interrogation regime was introduced, and the environment was understood by its designers as the foundation on which the questioning would build.

The Interrogation Program

The purpose of the black sites was interrogation, and the interrogation methods used at them were the program’s moral core and its ultimate undoing. The CIA had no interrogation doctrine of its own suited to long-term detention, so in 2002 it turned to two contract psychologists, James Mitchell and Bruce Jessen, who had worked with the military’s Survival, Evasion, Resistance, and Escape program. The SERE program trained American personnel to withstand the abusive interrogation methods used by hostile regimes, and it did so by exposing trainees, under controlled conditions, to those methods, including a simulated drowning technique. Mitchell and Jessen proposed to reverse-engineer the SERE curriculum, to take the methods designed to be resisted and apply them as an offensive interrogation system. Neither man had conducted a real interrogation of a hostile detainee. Neither had specialized expertise in al-Qaeda or in the languages and cultures of the men they would interrogate. They were hired anyway, and the company they later formed was paid tens of millions of dollars for its work on the program.

The methods they devised were given the bureaucratic label of enhanced interrogation techniques, a phrase chosen specifically to avoid the word torture. The techniques included walling, in which a detainee was slammed repeatedly against a wall; stress positions held for extended periods; sleep deprivation that in some cases extended past one hundred and eighty hours, achieved by shackling detainees in standing positions; confinement in coffin-sized and smaller boxes; dietary manipulation; nudity; exposure to cold; and the deliberate exploitation of phobias. The most notorious technique was waterboarding, in which a detainee was strapped to an inclined board, his face covered with a cloth, and water poured over the cloth to produce the sensation and physiological reality of drowning. Abu Zubaydah was waterboarded eighty-three times in August 2002. Khalid Sheikh Mohammed was waterboarded one hundred and eighty-three times in March 2003. The numbers are precise because they appear in the CIA’s own contemporaneous records, records the agency kept and that later investigators read.

Black sites also produced abuses that exceeded even the authorized techniques, because once a facility is operating in secret and its staff understand that the rules have been suspended, the rules tend to keep moving. The Senate investigation documented detainees subjected to what the agency itself called rectal rehydration and rectal feeding, procedures with no medical justification that functioned as a means of control and humiliation. It documented detainees threatened with harm to their families, including threats to harm their children and sexually assault their mothers. It documented mock executions. It documented the case of Gul Rahman, who froze to death in the Salt Pit. It documented detainees who suffered hallucinations, self-harm, and lasting psychological damage. The line between the authorized program and unauthorized abuse, which the program’s defenders insisted was firm, was in practice a line that the secrecy of the sites made unenforceable.

The role of medical personnel in the interrogation regime drew particular criticism after the fact. Physicians and psychologists were present during the application of the harshest techniques, and their presence was justified internally as a safeguard, a means of ensuring that interrogations did not cross from severe pressure into lethal injury. Critics, including medical ethicists and several professional bodies, argued that this role inverted the purpose of medicine. A doctor monitoring a waterboarding session to determine how much a detainee could endure before the technique had to stop was not protecting the detainee. The doctor was calibrating the abuse, lending it a veneer of clinical control, and making it possible to continue. The participation of health professionals in the interrogation regime became one of the most disputed legacies within those professions, and it illustrated a wider truth about the enterprise, which was that it drew respected institutions and credentialed experts into its work and, in doing so, implicated them in it.

The interrogation program rested on a theory, advanced by Mitchell and Jessen and accepted by the agency’s leadership, that severe physical and psychological pressure would induce in a detainee a state of learned helplessness, breaking his will to resist and rendering him compliant and truthful. The theory was never validated by evidence before it was adopted, and the Senate investigation concluded that the evidence assembled afterward did not support it. The CIA’s medical and psychological personnel raised concerns at various points. Some interrogators objected. The FBI, whose own interrogators used rapport-based methods and had achieved results with Abu Zubaydah before the enhanced techniques began, pulled its personnel out of the program rather than participate. The internal dissent is important, because the later defence of the program would describe it as the consensus response of a frightened government, when in fact it was contested inside the government from the beginning by professionals who believed both that it was wrong and that it would not work.

Exposure: How the Secret Came Apart

A program built on deniability survives only as long as its secrets hold, and the secrets of extraordinary rendition began leaking almost as soon as the program reached full scale. The unravelling did not come from a single dramatic disclosure. It came from the accumulation of dozens of separate threads, pulled by journalists, lawyers, parliamentarians, prosecutors, and the program’s own former participants, until the whole fabric came apart. The process took roughly a decade, and tracing it shows precisely how a covert operation conducted by a democracy is eventually forced into the light.

The first threads were journalistic. In 2002 and 2003, reporters began describing a system of secret detention and the transfer of prisoners to countries that practiced torture. In November 2005, The Washington Post published an account of the CIA black site network, and around the same time the broadcaster ABC News and other outlets added detail. The reporting did not name every site, but it established the existence of the archipelago, and it forced the issue into the open. Investigators outside the United States then took over the work. Human rights organizations and journalists obtained and analyzed thousands of aircraft flight records, cross-referencing tail numbers, corporate registrations, airport handling charges, and crew hotel bookings. The flight-log analysis was painstaking and devastating, because it converted a deniable program into a map. A plane that left a military base in the United States, stopped at a European airport, continued to the airfield nearest a suspected prison, and returned, leaving fuel receipts and landing fees at every stop, is not a deniable plane.

European institutions then escalated. The Council of Europe appointed the Swiss senator Dick Marty to investigate, and his reports, issued in 2006 and 2007, concluded that numerous European states had cooperated with the program, hosting sites or permitting transfer flights through their airspace and airports. The European Parliament conducted its own inquiry through a temporary committee, which reached similar conclusions and criticized member governments for their cooperation and their subsequent lack of candour. National parliaments in Poland, Romania, Lithuania, Italy, Germany, and the United Kingdom opened their own investigations. The program had been designed to keep American hands clean by spreading the work across many countries, and the design backfired completely, because every one of those countries had its own free press, its own opposition politicians, its own prosecutors, and its own courts, and each of those institutions had every incentive to expose what its government had concealed.

The role of leaked documents and of former participants who chose to speak should not be understated, because an operation that depends on the silence of everyone who touches it is undone the moment any one of them talks. Some of the most detailed disclosures came from officials inside or recently departed from the system who concluded that the public had a right to know what had been done in its name. Court proceedings in several countries forced the release of cables, memoranda, and operational records that the agency would never have disclosed voluntarily. Detainees who survived the system and were eventually released, men like Khaled El-Masri and Binyam Mohamed, gave detailed accounts to lawyers and journalists, and those accounts could be cross-checked against the flight logs and the host-country records until a consistent picture emerged. The architects had assumed that the prisoners would never be in a position to testify and that the officers would never break ranks. Both assumptions proved false, and the combination of surviving witnesses, departing insiders, and litigation-forced disclosure meant that the secret, once it began to leak, could not be resealed.

The most consequential thread inside the United States was the destruction of evidence, which became a scandal in its own right and which paradoxically guaranteed a deeper investigation. In 2005 the CIA destroyed ninety-two videotapes that recorded the interrogations of Abu Zubaydah and Abd al-Rahim al-Nashiri, including the waterboarding sessions. The destruction was ordered by José Rodriguez, then the head of the agency’s clandestine service. When the destruction became public in 2007, it triggered a criminal inquiry and, more importantly, it pushed the Senate Select Committee on Intelligence to launch a comprehensive review of the entire detention and interrogation program. That review, begun in 2009, examined more than six million internal CIA documents over several years. Its declassified summary, released on December 9, 2014, ran to roughly 525 pages and reached two central conclusions. The first was that the program had been far more brutal than the agency had represented to Congress, the White House, and the public. The second was that the program had not produced unique, otherwise-unavailable intelligence that disrupted plots or saved lives. The CIA disputed the second conclusion vigorously. It did not seriously dispute the first.

By the time the Senate summary appeared, the program itself had already been formally ended. On January 22, 2009, two days after taking office, President Barack Obama signed an executive order directing the CIA to close any detention facilities it operated and prohibiting it from maintaining such facilities in the future, and another order limiting interrogation to the methods in the Army Field Manual. The black sites were closed. The enhanced techniques were prohibited. The program that had been built in an atmosphere of emergency was dismantled by executive signature, and its dismantling is itself a data point, because a method that a successor administration eliminates within forty-eight hours is a method that had become a strategic liability rather than a strategic asset. The exposure had done its work. The fiction of deniability was finished, and what remained was the accounting.

The Legal Reckoning

The legal consequences of extraordinary rendition unfolded across two continents and many years, and the pattern they revealed is the single most important lesson of the entire episode for any state contemplating covert action. Inside the United States, the courts largely shielded the program from civil liability. Outside the United States, they did not, and the foreign judgments converted a secret program into a matter of established legal record.

Inside the United States, victims who tried to sue ran into the state secrets privilege, a doctrine that allows the government to ask a court to dismiss a case on the ground that litigating it would expose information damaging to national security. Khaled El-Masri sued, and his case was dismissed on state secrets grounds, a dismissal upheld on appeal, leaving the German car salesman with no American remedy for his abduction and abuse. Maher Arar sued, and his case too was dismissed by the American courts, even as Canada conducted a full public inquiry, cleared him completely, issued a formal apology, and paid him compensation of roughly ten and a half million Canadian dollars. A group of former detainees brought a case against the contractor company that arranged transport, Jeppesen Dataplan, a Boeing subsidiary, and that case was also dismissed on state secrets grounds. The American legal system, in other words, declined to adjudicate the program’s lawfulness, and the practical effect was impunity at home for the institutions and officials involved. No senior official was prosecuted. The only American imprisoned in connection with the program was a former CIA officer, John Kiriakou, who was convicted of disclosing the identity of a covert colleague after he spoke publicly about waterboarding.

Outside the United States, the picture was the reverse, and the contrast is the heart of the matter. The European Court of Human Rights, ruling on cases brought against the host and transit states, found repeatedly that the program had violated the European Convention on Human Rights. In the case of El-Masri against Macedonia, decided by the Grand Chamber in 2012, the court found that Macedonia had been responsible for his unlawful detention and ill-treatment and for handing him to the CIA, and the judgment described what was done to him at Skopje airport as treatment amounting to torture. In cases concerning Poland, decided in 2014, the court found that Poland had knowingly hosted a CIA black site and was responsible for the treatment of Abu Zubaydah and al-Nashiri there. In the case concerning Italy, decided in 2016, the court found Italy responsible in connection with the Abu Omar kidnapping. Romania and Lithuania were the subject of later adverse rulings. Each judgment ordered the responsible state to pay compensation, and each judgment placed on the permanent public record a judicial finding that the program had been unlawful.

The Italian criminal courts went further than any other jurisdiction. Following the prosecution built by Armando Spataro from the Milan kidnapping evidence, an Italian court in 2009 convicted twenty-three Americans, most of them CIA officers, in absentia for the abduction of Abu Omar. They were sentenced to prison terms and could not safely travel to countries that might extradite them to Italy. One of them, a former CIA officer named Sabrina De Sousa, was later detained in Europe and faced the prospect of serving the Italian sentence before the matter was resolved through legal and diplomatic channels. The Italian convictions were the first and remain among the very few criminal verdicts arising anywhere from the program, and they demonstrated a principle that intelligence services prefer to ignore, which is that an officer who conducts a covert operation on the soil of a functioning legal democracy can be tried by that democracy’s courts, named in its judgments, and rendered a fugitive for the rest of his career.

Civil litigation in the United States deserves a closer look, because the state secrets doctrine that defeated it is itself a revealing piece of the story. The doctrine allows the executive branch to ask a court to dismiss a lawsuit, before any evidence is heard, on the assertion that the litigation would expose information harmful to national security. In the rendition cases the government deployed the doctrine repeatedly and successfully, and the courts, with rare exceptions, accepted it. The practical result was that the question of whether the program had been lawful was never adjudicated on its merits by an American court. A man could be abducted in error, abused, and held for months, and when he sought a remedy in the courts of the country responsible, he would be told that the very facts of his case were too secret to be examined. The doctrine protected the program from domestic legal accountability, but it did so at a cost that is easy to overlook. It signaled that the American legal system, faced with credible allegations of grave abuse by its own intelligence service, would decline to look, and that signal did its own quiet damage to the country’s standing as a society governed by law rather than by executive assertion.

The legal reckoning therefore produced a split outcome. The state that ran the program protected its own officials from prosecution and its own institutions from civil liability through the state secrets doctrine. The states that had cooperated, the smaller partners who had hosted the sites and permitted the flights, absorbed the judicial findings, paid the compensation, and carried the political damage. The covert operative’s nightmare is not abstract censure. It is a named conviction, a sentence, a court judgment with his role in it, and a passport that has become a liability. Extraordinary rendition produced all of those, and produced them in volume, and any intelligence service studying the episode would have read the legal record as a detailed manual of what not to build.

Key Figures

The program was not an abstraction. It was designed, authorized, operated, and ultimately exposed by identifiable people, and a handful of them illustrate the institutional dynamics that drove the whole enterprise.

George Tenet and the Agency Leadership

George Tenet was the Director of Central Intelligence from 1997 to 2004, which means he led the CIA through the September 11 attacks and through the construction of the detention and interrogation program. Tenet presided over an agency that had been criticized for failing to prevent the attacks and that was now determined never to be caught unprepared again. He sought and obtained legal authorizations for the program, pressed the Justice Department for written assurances that the methods were lawful, and defended the program after his retirement, arguing in his memoir and in interviews that the interrogations had produced intelligence that saved lives. Tenet embodies the leadership psychology that produced the program, the conviction that the institution’s survival and the nation’s safety required methods that the institution would previously have rejected, combined with a careful insistence on obtaining paper authorization for each step so that responsibility could later be shown to lie elsewhere.

Cofer Black and the Counterterrorist Center

Cofer Black ran the CIA’s Counterterrorist Center at the time of the September 11 attacks and in the immediate aftermath. It was Black who told a congressional committee that after September 11 the gloves had come off, and the phrase remains the most honest single description of the operational culture that produced extraordinary rendition. Black’s center was the bureaucratic engine of the program in its early phase, the unit that assessed targets, planned captures, and ran the first black sites. He represents the operational tier, the officials who translated the leadership’s authorizations and the lawyers’ memoranda into actual seizures and actual cells.

James Mitchell and Bruce Jessen, the Contract Psychologists

James Mitchell and Bruce Jessen were the two psychologists, formerly associated with the military SERE program, who designed and in important cases personally administered the enhanced interrogation techniques. Neither had interrogated a hostile detainee before the program began. Their company received tens of millions of dollars in CIA contracts. They are the clearest example of a recurring feature of the program, the outsourcing of its most sensitive and most legally exposed work to private contractors who operated at the edge of, and sometimes outside, the accountability structures that governed government employees. A lawsuit brought against them by former detainees, unlike the suits against the agency itself, was not dismissed on state secrets grounds and was eventually settled.

José Rodriguez and the Destroyed Tapes

José Rodriguez was the head of the CIA’s clandestine service who, in 2005, ordered the destruction of the ninety-two videotapes recording the early interrogations. Rodriguez defended the destruction as the elimination of material that endangered the officers depicted, and he defended the program as a whole in a combative memoir. His decision to destroy the tapes is one of the program’s defining moments, because it was both an attempt to bury evidence and the act that, once exposed, made a far more thorough investigation inevitable. Rodriguez represents the instinct toward concealment that runs through every covert program and the way that instinct, acted upon, tends to deepen rather than contain the eventual scandal.

Khaled El-Masri, Maher Arar, and Abu Omar, the Subjects

The men the program seized are figures in the story as much as the men who ran it. Khaled El-Masri, the German car salesman, became the face of mistaken identity and the litigant whose case the European Court of Human Rights used to establish the program’s unlawfulness. Maher Arar, the Canadian engineer, became the face of the airport interception and the subject of the most thorough official exoneration any rendition victim received. Hassan Mustafa Osama Nasr, the Milan cleric known as Abu Omar, became the subject of the only substantial criminal convictions the program produced. Their names matter because a covert program designed to be anonymous was, in the end, defined publicly by the named individuals it wronged, and because the gap between the program’s intended invisibility and the visibility of these three men is the gap that destroyed it.

The Investigators: Dick Marty, Armando Spataro, and the Senate Committee

The enterprise was exposed by people as identifiable as the people who ran it, and three of them stand for the wider effort. Dick Marty, the Swiss senator appointed by the Council of Europe, conducted the inquiry that mapped European complicity and named the cooperating states, producing reports that made it impossible for European governments to maintain their denials. Armando Spataro, the Milan prosecutor, built the criminal case from the Abu Omar kidnapping with ordinary investigative tools, mobile phone records, hotel registrations, and patient cross-referencing, and secured the only substantial criminal convictions the operation produced. The Senate Select Committee on Intelligence, working under the chairmanship of Senator Dianne Feinstein, conducted the multi-year review of six million internal documents that produced the definitive account of the brutality involved and the definitive challenge to the claimed effectiveness. These figures matter because they demonstrate the mechanism by which a covert operation in a democracy is brought to account. It is not a single heroic disclosure. It is the slow, institutional work of prosecutors, parliamentary investigators, and oversight committees, each pulling its own thread, until what was designed to be invisible has been documented in exhaustive public detail.

Consequences and Impact

The consequences of extraordinary rendition extended well beyond the courtrooms, and they fell into several distinct categories, each of which carried a cost that the program’s designers had not adequately weighed.

The first cost was strategic and reputational. The United States had spent the second half of the twentieth century positioning itself as the leading advocate of human rights and the rule of law, and it had built a great deal of its global influence on the credibility of that position. The disclosure that it operated secret prisons and tortured detainees damaged that credibility in a way that was difficult to repair, because the damage was not a matter of accusation but of documented fact, confirmed by its own Senate. Authoritarian states that the United States criticized for abuses now had a ready answer. Allies who cooperated on counter-terrorism found that cooperation politically costly at home. The program had been justified as a measure to protect the country, and it produced a strategic wound that no plot disruption could offset.

A second cost was the radicalization effect. The images and accounts that emerged from the program, particularly when combined with the abuses at the Abu Ghraib military prison in Iraq, became recruitment material for the very organizations the program was meant to defeat. It is impossible to quantify precisely how many people were drawn toward militancy by the perception that the United States tortured Muslim detainees in secret prisons, but intelligence analysts and counter-terrorism scholars widely accepted that the effect was real and significant. A program designed to degrade terrorist organizations handed those organizations a propaganda asset of lasting value.

The third cost fell on the cooperating states and on the wider system of intelligence alliances. Poland, Romania, Lithuania, Italy, and others paid a domestic political price, faced adverse court rulings, and in some cases paid compensation. The episode made future intelligence cooperation more fraught, because any government asked to assist a covert American operation now had the rendition precedent in mind and understood that assistance offered in secret could become a public scandal years later. Trust within alliances is a finite resource, and the program spent a great deal of it.

The fourth cost was institutional and internal. The program consumed enormous resources, generated years of investigations, lawsuits, and congressional conflict, and left the CIA itself divided and damaged. The dispute between the agency and the Senate committee over the investigation became openly hostile, including a confrontation over whether agency personnel had improperly accessed computers used by Senate staff. An institution that depends on congressional trust and congressional funding had badly damaged its relationship with its principal oversight body. The program also left a generation of officers exposed, some of them named in foreign judgments, some of them carrying the personal and legal weight of having participated in operations that their own government’s later assessment condemned.

Set against these costs, the program’s defenders placed its claimed intelligence value, and the weighing of that claim is the subject of the debate that has never been fully resolved. What is not in dispute is that the costs were large, were lasting, and were borne not only by the program’s victims but by the country that ran it.

The Effectiveness Debate

The central analytical question about extraordinary rendition is whether it worked, and the question has two genuine sides, which is why it remains contested rather than settled. The debate matters far beyond the American case, because every state weighing a coercive counter-terrorism method weighs it against an implicit claim of effectiveness, and the rendition program is the most thoroughly documented test of that claim available.

Defenders made the case that the program worked most forcefully through the CIA itself and through senior officials including George Tenet and the agency directors who followed him. Their argument ran as follows. The men captured and interrogated were among the most dangerous figures in al-Qaeda, including the principal organizer of the September 11 attacks. The interrogations, including the coercive interrogations, produced a large volume of intelligence about the organization’s structure, personnel, plans, and methods. Some of that intelligence, the agency argued, contributed to the disruption of specific plots and to the identification and capture of other operatives. The agency contended that certain detainees provided information only after the enhanced techniques were applied, and that this information could not have been obtained otherwise. In the agency’s telling, the program was a grim necessity that, whatever its moral cost, prevented attacks and saved lives.

The case that the program did not work was made most comprehensively by the Senate Select Committee on Intelligence, whose multi-year review of six million internal documents concluded that the CIA’s effectiveness claims were inflated and in important instances false. The committee’s analysis examined the specific plots and captures the agency cited as successes and concluded that the relevant intelligence had, in case after case, either come from sources other than coercive interrogation, or been available before the coercion began, or been corroborative of information already held rather than uniquely decisive. The committee found that detainees subjected to the harshest treatment sometimes produced fabricated information, telling interrogators what they believed would end the abuse, and that the fabrications then had to be chased down and discounted. The FBI’s experience with Abu Zubaydah, in which rapport-based questioning produced cooperation before the enhanced techniques were introduced, was offered as evidence that the lawful alternative was not only available but more reliable. The committee’s bottom line was that the program had not been the indispensable tool the agency claimed, and that its costs had therefore been incurred for a benefit far smaller than advertised.

No single case illustrates the dispute better than the trail that eventually led to Osama bin Laden’s compound in Abbottabad. The agency’s defenders pointed to that hunt as the strongest vindication available, arguing that detainees in CIA custody had supplied the early identification of the courier whose movements were eventually tracked to the house. The Senate reviewers examined the same record and reached a sharply different reading. They found that the courier’s significance had been flagged by a detainee who was questioned before coercive techniques were applied to him, that other detainees subjected to the harshest treatment had actively understated the courier’s importance, and that the decisive identification work had been done through years of patient analytic and signals effort rather than through anything extracted in a cell. The reconstruction of that decade-long manhunt, traced in our account of the CIA hunt for bin Laden, shows a result produced by traditional tradecraft that the coercive enterprise then tried to claim as its own. When the single best example offered for a method is contested by the government’s own most thorough investigation, the example has stopped functioning as proof and has become instead a measure of how far the original claims had drifted from the documented facts.

The professional interrogation literature largely supports the skeptical side. A long tradition of research and operational experience holds that torture and severe coercion are poor methods for obtaining accurate information, because a person in extreme distress is motivated to make the distress stop, not to tell the truth, and will say whatever he believes the interrogator wants. Rapport-based methods, which build a relationship and exploit a detainee’s own psychology, have a stronger track record. This is not a sentimental position. It is an operational one, advanced by experienced interrogators who object to coercion partly because it does not reliably work. The honest analytical conclusion is that the effectiveness question cannot be resolved with certainty, because the counter-factual, what these detainees would have revealed under lawful interrogation, can never be fully tested. But the weight of the documented evidence, including the CIA’s own records as read by the Senate, runs against the claim that the program produced unique and otherwise-unavailable intelligence. The program’s defenders are left arguing necessity for a method whose necessity its own government’s most thorough investigation could not confirm.

For any state studying the episode, the effectiveness debate carries a specific warning. A coercive program imposes certain, large, and lasting costs, the legal exposure, the reputational damage, the radicalization effect, in exchange for an intelligence benefit that is uncertain and contested. A rational strategist weighing certain costs against uncertain benefits will be cautious, and the rendition program is the case study that makes that calculation concrete. This is the lens through which the architects of other counter-terrorism doctrines, including those examined in our study of Israel’s Rise and Kill First targeted-killing doctrine, assessed the American experience and drew their own conclusions about what to copy and what to avoid.

Why It Still Matters

Extraordinary rendition ended as an active program more than a decade ago, but it continues to matter because it functions as a controlled experiment in how a democracy conducts covert counter-terrorism, and the results of that experiment shaped the choices of states that came afterward. The most instructive of those states, for an Indian readership, is India itself, and the comparison between the American model and the Indian one is the reason this history belongs in any serious analysis of the shadow war.

Begin with what the rendition program got wrong, mechanically. It built a system. It created institutions, the black sites, the contractor companies, the transport fleet, and institutions generate records, employ people, and answer to oversight bodies. It depended on the cooperation of a dozen foreign governments, and each of those governments was a separate point of failure with its own free press and its own courts. It kept meticulous internal documentation, including videotapes and cables, on the theory that the records were secret, when in fact the records became the evidence. It captured rather than killed, which meant it accumulated living witnesses, men like El-Masri and Arar and Abu Omar who survived to testify, to sue, and to give the program a human face in courtrooms across Europe. And it operated on the soil of functioning legal democracies, which meant that prosecutors like Armando Spataro could build criminal cases against its officers. Every one of these features was a vulnerability, and together they guaranteed exposure.

Now consider the method India is widely assessed to have adopted in its campaign against terrorists sheltered in Pakistan, the campaign analyzed in full in our overview of India’s shadow war against terror. The shadow war, as the open-source pattern suggests it, does not build a system of the rendition kind. There is no archipelago of secret prisons, because there are no prisoners. There are no transport aircraft with traceable tail numbers, no contractor companies on a payroll, no host governments to be made complicit, because the operations as reported take place on the target’s own soil and require no foreign cell. There are no interrogation videotapes, because there are no interrogations. There are, above all, no living witnesses of the El-Masri kind, because the method as observed is killing rather than capture, and the dead do not sue, do not testify, and do not give the operation a face in a European courtroom. Where the American program left a paper trail that filled six million documents, the shadow war as described leaves a series of unattributed deaths and an official posture of complete denial.

The contrast is the lesson, and it is a lesson about deniability. The United States wanted its program to be deniable and designed it to be deniable, and it failed, because capture, detention, and interrogation are activities that inherently generate records, witnesses, and complicit partners. India, if the prevailing analysis is correct, appears to have understood that the rendition model is structurally undeniable and to have chosen a method that is structurally deniable instead. Kill rather than capture, so there are no prisoners. Operate alone on the target’s own ground, so there are no partners. Keep no facilities, so there is nothing to discover. Deny rather than explain, so there is no official narrative to be falsified. This is not a claim that the Indian method is morally superior. The legal and ethical objections to extrajudicial killing are at least as serious as the objections to rendition, and our examination of the targeted-killing legal and ethical debate lays them out in detail. It is a claim that the Indian method is operationally more survivable, because it produces far less of the evidence that destroyed the American program.

There is a harder edge to this contrast that the survivability framing can obscure, and it deserves to be stated plainly. The rendition system, for all its abuses, retained one feature that an elimination method discards entirely, which is the bare possibility of correcting a mistake. Khaled El-Masri was the wrong man, and because he had been seized rather than killed he was eventually able to walk out of a cell, find a lawyer, and force a continent to examine what had been done to him. His survival is the only reason his case became a lesson at all. A method built on killing offers no such opening. When the target on the far side of a denied operation turns out to have been misidentified, there is no release, no lawsuit, and no correction, because the error is buried with the body and the official posture of denial ensures that no inquiry is ever opened. So the comparison cuts two ways. Capture-based coercion is operationally fragile precisely because it leaves survivors, and elimination-based action is operationally durable precisely because it does not, which means the more survivable method is also the one that makes its own worst mistakes permanent and invisible. That is not an argument for rendition. It is a reminder that operational survivability and moral acceptability are different axes, and that a method can score well on the first while failing catastrophically on the second.

The rendition program also matters because it sits inside a larger pattern that recurs across the history of covert state action. Time and again, a state under threat adopts a coercive secret method, convinces itself that the emergency justifies it, operates it for a period in the belief that it will stay hidden, and then watches it surface, because secrecy degrades and democracies in particular contain institutions whose function is to expose. The pattern is visible in the American drone campaign, compared directly with the Indian approach in our analysis of the US drone program measured against India’s shadow war. It is visible in the Israeli campaigns reconstructed in our account of Mossad’s Operation Wrath of God. It is visible, in its darkest form, in the apartheid-era death squads. Extraordinary rendition is one node in that pattern, and its particular contribution is the clarity of its failure. It is the case where the gap between intended deniability and actual exposure is widest, and it is therefore the case that taught the most.

The final reason the program still matters is that it poses a question no democracy has answered well, which is whether a state bound by law can conduct effective covert counter-terrorism at all, or whether the attempt always forces a choice between a method that respects the law and is therefore limited and a method that is effective and therefore lawless. The United States tried to have both. It tried to operate outside the law while maintaining the appearance of operating within it, and the appearance could not be sustained. The states that watched did not solve the underlying problem. They simply chose their side of it more decisively, some by acknowledging their methods and defending them openly, others by adopting methods so deniable that the question of acknowledgement need never arise. Extraordinary rendition is the program that made the choice unavoidable, and that is why, long after the last black site was closed, the history of how the CIA kidnapped terror suspects across the globe remains required reading for anyone trying to understand how states fight their shadow wars and what those wars cost.

Frequently Asked Questions

Q: What was the US extraordinary rendition program?

The US extraordinary rendition program was a covert operation, run primarily by the CIA after the September 2001 attacks, in which terrorism suspects were seized in foreign countries without arrest warrants or extradition proceedings and transferred either to secret CIA prisons known as black sites or to the custody of foreign governments that practiced torture. The word rendition refers to the transfer of a person from one jurisdiction to another. What made this version extraordinary was the removal of any court process from the procedure. Suspects were not brought to trial. They were taken to cells. The program was authorized by a classified presidential directive signed on September 17, 2001, and it was designed around deniability, with the United States intending to keep its own hands officially clean while the worst treatment occurred either at hidden sites or in the prisons of cooperating states.

Q: How did the CIA kidnap terror suspects?

The CIA used three main capture methods. The first was the street snatch, in which a team seized a suspect in a public place, as happened to the cleric Abu Omar in Milan in February 2003. The second was the airport interception, in which a suspect was detained at a border crossing he had a legal right to use, as happened to the Canadian engineer Maher Arar at New York’s Kennedy airport in September 2002. The third and most common was the handover, in which a foreign government, often Pakistan, detained a suspect and delivered him to American officers, frequently at an air base with little documentation. After capture, the suspect was photographed naked, searched, shackled, hooded, fitted with a diaper, and chained to the floor of an aircraft for transfer to a black site or to a third country.

Q: Where were the CIA’s black sites?

The CIA operated a scattered network of secret prisons that investigators came to call the black site archipelago. Through journalism, leaked documents, and rulings of the European Court of Human Rights, the sites were located in several countries. An early facility operated in Thailand, where Abu Zubaydah was held in 2002. A site in Afghanistan, known to detainees as the Salt Pit, was where the prisoner Gul Rahman froze to death in November 2002. Sites operated in Poland and Romania, in former military or intelligence compounds, holding high-value detainees in the program’s later years, and a site operated for a period in Lithuania. Detainees were also held at facilities connected to other cooperating states. The CIA placed the prisons abroad specifically to keep them off American soil and therefore beyond the reach of American courts.

Q: Was torture used in the rendition program?

Yes. The CIA used what it called enhanced interrogation techniques, a phrase chosen specifically to avoid the word torture. The techniques included waterboarding, in which a detainee experienced controlled drowning, as well as sleep deprivation extending past one hundred and eighty hours, walling, stress positions, confinement in small boxes, nudity, cold exposure, and dietary manipulation. The CIA’s own records show Abu Zubaydah was waterboarded eighty-three times and Khalid Sheikh Mohammed one hundred and eighty-three times. The 2014 Senate investigation also documented abuses beyond the authorized techniques, including rectal feeding without medical justification, threats against detainees’ families, and mock executions. The European Court of Human Rights described the treatment of Khaled El-Masri at Skopje airport as amounting to torture. The program also caused at least one death in custody.

Q: Did rendition produce useful intelligence?

This is the most contested question about the program. The CIA argued that the interrogations, including the coercive ones, produced valuable intelligence about al-Qaeda’s structure and plans and contributed to the disruption of plots. The Senate Select Committee on Intelligence, after reviewing more than six million internal CIA documents, reached the opposite conclusion, finding that the agency’s effectiveness claims were inflated and that the intelligence cited as program successes had generally come from other sources, been available before coercion began, or merely corroborated information already held. Professional interrogators widely hold that coercion produces unreliable information, because a person in severe distress will say anything to make it stop. The honest assessment is that the effectiveness claim cannot be proven, and that the weight of the documented evidence runs against it.

Q: What legal consequences did the program face?

The legal consequences split sharply by jurisdiction. Inside the United States, victims who sued, including Khaled El-Masri and Maher Arar, had their cases dismissed under the state secrets privilege, and no senior official was prosecuted. Outside the United States, the picture reversed. The European Court of Human Rights ruled repeatedly that the program violated the European Convention, finding Macedonia, Poland, Italy, Romania, and Lithuania responsible in connection with specific cases and ordering compensation. Italian criminal courts convicted twenty-three Americans, most of them CIA officers, in absentia in 2009 for the Milan kidnapping. Canada conducted a full public inquiry into the Arar case, exonerated him, apologized, and paid compensation of roughly ten and a half million Canadian dollars.

Q: Why did India choose killing over capture?

India is widely assessed to have studied coercive Western counter-terrorism programs, including extraordinary rendition, and to have concluded that capture-based methods are structurally undeniable. Capture produces prisoners, prisons, transport, complicit partners, interrogation records, and living witnesses, and every one of those generates evidence. The rendition program left a trail of six million documents, dozens of court judgments, and victims who testified across Europe. A method based on killing rather than capture produces none of that. There are no prisoners to be discovered, no facilities to be exposed, no witnesses to sue or testify. The reported pattern of the shadow war suggests India chose the deniable method precisely because the American experience demonstrated that the rendition model cannot stay secret.

Q: How does rendition compare to targeted killing?

Rendition and targeted killing are both extrajudicial counter-terrorism methods, but they differ in a way that matters enormously for deniability. Rendition is capture-based. It keeps the suspect alive, which means the operating state accumulates prisoners, must hold them somewhere, must interrogate them, and must rely on partner countries, all of which generate records and witnesses. Targeted killing is elimination-based. It produces a death and an unattributed crime scene, but no prisoner, no prison, no interrogation, and no surviving witness who can testify to what was done. Targeted killing raises legal and ethical objections at least as severe as rendition, including the absence of any possibility of error correction once the target is dead. But as a matter of operational survivability, the killing method leaves far less of the evidence that exposed the rendition program.

Q: When did extraordinary rendition begin?

A narrow form of rendition existed long before the September 2001 attacks. American agencies had used rendition for decades to bring fugitives back for trial, and under President Bill Clinton in the mid-1990s the CIA ran a rendition program that transferred suspected militants, mostly to Egypt, where they faced existing legal proceedings, however flawed. Roughly seventy such transfers took place before 2001. The crucial change came after the September 2001 attacks, when the September 17 presidential directive authorized the CIA to capture and detain suspects on its own authority. The destination shifted from a courtroom to a cell, and the practice expanded into the global black site and interrogation enterprise that the word extraordinary rendition now describes.

Q: What is the difference between rendition and extraordinary rendition?

Ordinary rendition is the transfer of a suspect from one jurisdiction to another to face a legal process, typically a criminal trial. It bypasses formal extradition but still ends in a courtroom. Extraordinary rendition removes the courtroom entirely. The suspect is transferred not to face trial but to be detained and interrogated, either at a secret CIA prison or in the custody of a foreign government chosen because its security services practiced torture. The defining feature of the extraordinary version is the absence of any judicial process at any stage, no warrant at the start and no trial at the end, which is what placed it outside the rule of law rather than merely at the edge of it.

Q: Who were James Mitchell and Bruce Jessen?

James Mitchell and Bruce Jessen were two psychologists who designed the CIA’s enhanced interrogation program. Both had worked with the military’s Survival, Evasion, Resistance, and Escape training, which teaches American personnel to withstand abusive interrogation by exposing them, under controlled conditions, to such methods. Mitchell and Jessen proposed to reverse-engineer that training, turning resistance techniques into an offensive interrogation system. Neither had ever interrogated a hostile detainee before the program began, and neither had specialized expertise in al-Qaeda. The company they later formed received tens of millions of dollars in CIA contracts. They are the clearest example of how the program outsourced its most legally exposed work to private contractors.

Q: What did the 2014 Senate report conclude?

The Senate Select Committee on Intelligence released the declassified summary of its review on December 9, 2014. The summary ran to roughly 525 pages and was based on the examination of more than six million internal CIA documents. It reached two central conclusions. First, the program had been substantially more brutal than the CIA had represented to Congress, the White House, and the public, and it documented specific abuses including a death in custody and treatment beyond the authorized techniques. Second, the program had not produced the unique, otherwise-unavailable intelligence the agency claimed, and the effectiveness cases the CIA cited did not hold up under examination. The CIA disputed the second conclusion but did not seriously contest the first.

Q: What happened to Khaled El-Masri?

Khaled El-Masri was a German citizen and car salesman seized at the Macedonia-Serbia border on December 31, 2003, because his surname resembled that of a genuine suspect. He was held in a Skopje hotel for twenty-three days, then handed to a CIA team, abused, and flown to the Salt Pit prison in Afghanistan, where he was held for about four months. The CIA realized within weeks that it had the wrong man but argued internally for months before releasing him on a mountain road in Albania. His American lawsuit was dismissed under the state secrets privilege. In 2012 the Grand Chamber of the European Court of Human Rights ruled that Macedonia was responsible for his unlawful detention and ill-treatment, a landmark judgment establishing the program’s unlawfulness.

Q: Who was Maher Arar and what happened to him?

Maher Arar was a Canadian telecommunications engineer detained at New York’s Kennedy airport in September 2002 while returning home from a family holiday, on the basis of unreliable intelligence linking him loosely to a person of interest. Rather than deport him to Canada, where he was a citizen and resident, American officials flew him to Jordan and then to Syria, his country of birth, where he was held for nearly a year and beaten. Arar had no connection to terrorism. A Canadian public inquiry led by Justice Dennis O’Connor examined his case in detail, exonerated him completely, and led to a formal Canadian government apology and compensation of roughly ten and a half million Canadian dollars. His American lawsuit was dismissed by the courts.

Q: Did anyone face criminal prosecution for the rendition program?

Criminal accountability was almost entirely confined to foreign courts. Inside the United States, no senior official was prosecuted for designing or running the program. The only American imprisoned in connection with it was a former CIA officer, John Kiriakou, convicted of disclosing a covert colleague’s identity after speaking publicly about waterboarding, meaning he was punished for revealing the program rather than for conducting it. The substantial criminal verdicts came from Italy, where a court in 2009 convicted twenty-three Americans, most of them CIA officers, in absentia for the abduction of Abu Omar in Milan, leaving them as fugitives unable to travel to jurisdictions that might extradite them.

Q: What role did European governments play in the rendition program?

European governments were deeply involved, and their involvement became the program’s principal vulnerability. Poland, Romania, and Lithuania hosted secret CIA prisons on their territory. Italy permitted, through elements of its intelligence service, the kidnapping of Abu Omar from Milan. Many states allowed CIA transfer flights to cross their airspace and refuel at their airports. Investigations by the Council of Europe, led by senator Dick Marty, and by the European Parliament documented this cooperation. The European Court of Human Rights subsequently ruled against several of these states. Because the program depended on the silence of so many democracies, each with its own free press, courts, and parliament, it had many points of failure, and the cooperation that was meant to enable it ultimately exposed it.

Q: Did the rendition program officially end?

The program of secret CIA detention and enhanced interrogation was formally ended in January 2009. Two days after taking office, President Barack Obama signed an executive order directing the CIA to close any detention facilities it operated and prohibiting it from maintaining such facilities in the future, along with a further order limiting interrogation to the methods permitted in the Army Field Manual. The black sites were closed and the enhanced techniques prohibited. The speed of the dismantling, accomplished within forty-eight hours of a change of administration, is itself revealing, indicating that the program had become a strategic liability rather than an asset that a new government wished to retain.

Q: What does the rendition program reveal about democratic counter-terrorism?

The program functions as a controlled experiment in whether a law-bound democracy can run an effective covert counter-terrorism operation. The American attempt was to have it both ways, to operate outside the law while preserving the appearance of operating within it, and the appearance could not be sustained, because capture, detention, and interrogation inherently generate records, witnesses, and complicit partners, and because democracies contain free presses, courts, and oversight bodies whose function is exposure. The lesson other states drew was stark. A coercive program imposes certain and lasting costs in exchange for uncertain benefits, and a program built on captured prisoners cannot stay secret. Those conclusions shaped the more deniable methods that later campaigns, including India’s shadow war, were assessed to have adopted.

Q: What was the difference between a CIA black site and Guantanamo Bay?

The two were distinct kinds of facility serving distinct purposes within the same broad architecture. A black site was a covert prison whose existence was classified, whose location was concealed even from most of the United States government, and whose detainees were held entirely outside any acknowledged legal process, with no name on a roster and no access to a lawyer or the Red Cross. Guantanamo Bay, by contrast, was an acknowledged military detention facility whose existence was public, whose detainees were eventually identified, and which became the subject of repeated litigation in American courts. Detainees were often moved from the secrecy of a black site into Guantanamo once the agency had finished with them, which is part of how the secret system came to light. The black site was where the practice happened in darkness. Guantanamo was the more visible holding facility into which some of its subjects eventually surfaced.

Q: How many people were subjected to extraordinary rendition?

The exact total has never been fully established, which is itself a consequence of how the system was designed. The 2014 Senate review identified one hundred nineteen detainees held in CIA custody at black sites, and concluded that at least twenty-six of them had been wrongfully held, meaning they did not meet the standard for detention that the agency itself had set. That figure, however, counts only the secret CIA prisons. It does not include the larger and far less documented group of suspects who were seized and handed directly to the custody of foreign governments rather than to a CIA site, a category for which no reliable public count exists. Independent investigations by journalists and human rights organizations have placed the broader total considerably higher, but the secrecy that defined the practice has also ensured that the full number of people who passed through it remains genuinely unknown.

Q: What did the psychologists mean by learned helplessness?

Learned helplessness was a concept borrowed from experimental psychology and repurposed as the theoretical foundation of the interrogation method. In its original research context it described what happens when a subject is exposed to distress it cannot control or escape, until it stops attempting to resist at all and becomes passive even when escape later becomes possible. The contract psychologists who designed the enhanced techniques argued that inducing a comparable state in a detainee would break his will to withhold information and make him compliant with his questioners. Critics, including many professional interrogators, regarded this as both ethically indefensible and operationally mistaken, because a person reduced to passive helplessness is not a person reliably telling the truth, but a person who will say whatever is necessary to make the distress end. The concept lent a scientific vocabulary to a practice that the documented evidence suggests did not deliver what the vocabulary promised.

Q: Did extraordinary rendition make future terrorism more or less likely?

This cannot be measured with precision, but the weight of expert assessment holds that the costs to long-term security outran the benefits. The images and accounts that emerged from the black sites and from the foreign prisons became a recruiting argument for the very groups the program was meant to weaken, supplying a propaganda narrative of Western hypocrisy that extremist organizations used directly. The practice also damaged intelligence cooperation with allied democracies, whose courts and publics recoiled from involvement once the details surfaced, and it eroded the moral standing the United States relied on when asking other governments to observe legal limits. Against those lasting costs sat a set of intelligence gains that the government’s own most thorough investigation could not confirm as unique. A method that generates a durable propaganda weapon for the adversary while delivering contested operational returns is difficult to defend as a net contribution to security, and that judgment is part of why later campaigns moved away from it.