Why States Won’t Give Nuclear Weapons to Terrorists
[ Page 83-84 ]
We conclude that neither a terror group nor a state sponsor would remain anonymous after a nuclear terror attack. We draw this conclusion on the basis of four main andings. First, data on a decade of terrorist incidents reveal a strong positive relationship between the number of fatalities caused in a terror attack and the likelihood of attribution. Roughly three-quarters of the attacks that kill 100 people or more are traced back to the perpetrators. Second, attri- bution rates are far higher for attacks on the U.S. homeland or the territory of a major U.S. ally—97 percent (thirty-six of thirty-seven) for incidents that killed ten or more people. Third, tracing culpability from a guilty terrorist group back to its state sponsor is not likely to be difacult: few countries sponsor ter- rorism; few terrorist groups have state sponsors; each sponsored terror group has few sponsors (typically one); and only one country that sponsors terror- ism, Pakistan, has nuclear weapons or enough assile material to manufacture a weapon. In sum, attribution of nuclear terror incidents would be easier than is typically suggested, and passing weapons to terrorists would not offer coun- tries an escape from the constraints of deterrence.12
[ Page 85-86 ]
Perhaps the most important reason to doubt the nuclear-attack-by-proxy scenario is the likelihood that the ultimate source of the weapon might be discovered.15 One means of identifying the state source of a nuclear terrorist attack is through “nuclear forensics”—the use of a bomb’s isotopic angerprints to trace the assile material device back to the reactors, enrichment facilities, or uranium mines from which it was derived. In theory, the material that remains after an explosion can yield crucial information about its source: the ratio of uranium isotopes varies according to where the raw uranium was mined and how it was processed, and the composition of weapons-grade plutonium re- veals clues about the particular reactor used to produce it and how long the material spent in the reactor.16 The possibility that the covert plot could be discovered before being carried out also acts as a deterrent. For these and other reasons, some analysts argue that nuclear terrorism is unlikely.
[ Page 92-93 ]
Given the enormous risks involved, it is difacult to imagine a state’s leaders placing so much faith in a terrorist organization unless they already had a long-running, close, and trusting relationship with that group, and unless that group had repeatedly demonstrated its reliability, competence, and ability to maintain secrecy. Furthermore, leaders considering giving nuclear weapons to terror groups would need to and a group with the demonstrated capabil- ity to conduct complex operations across international borders.27 Many violent nonstate groups can plant roadside bombs or conduct small-scale ambushes against unsuspecting targets, but those relatively simple attacks do not imply an ability to conduct complex international operations involving training, travel, visas, anances, and secure communications.28 In short, both the com- plexities of the mission and the need for unwavering trust mean that a state seeking to orchestrate a nuclear attack by proxy would be limited to collabora- tions with well-established terrorist organizations with which it has existing relationships, simplifying the task of connecting terrorist perpetrators to their state sponsors.
[ Page 102 ]
There are at least five reasons, however, to expect that attributing a nuclear terrorist attack would be easier than attributing a conventional terrorist attack. First, no terrorism investigation in history has had the resources that would be deployed to investigating the source of a nuclear terror attack—particularly one against the United States or a U.S. ally. Rapidly attributing the attack would be critical, not merely as a arst step toward satisfying the rage of the victims but, more importantly, to determine whether additional nuclear attacks were imminent. The victim would use every resource at its disposal— money, threats, and force—to rapidly identify the source of the attack.47 If necessary, any investigation would go on for a long time; it would never “blow over” from the victim’s standpoint.
The second reason why attributing a nuclear terror attack would be easier than attributing a conventional terrorist attack is the level of international assistance the victim would likely receive from allies, neutrals, and even adversaries. An attack on the United States, for example, would likely trigger unprecedented intelligence cooperation from its allies, if for no other reason than the fear that subsequent attacks might target them. Perhaps more impor- tant, even adversaries of the United States—particularly those with access to assile materials—would have enormous incentives to quickly demonstrate their innocence. To avoid being accused of sponsoring or supporting the attack, and thus to avoid the wrath of the United States, these countries would likely go to great lengths to demonstrate that their weapons were accounted for, that their assile materials had different isotopic properties than the type used in the attack, and that they were sharing any information they had on the attack. The cooperation that the United States received from Iran and Pakistan in the wake of the September 11 attacks illustrates how potential adversaries may be motivated to help in the aftermath of an attack and stay off the target list for retaliation.48 The pressure to cooperate after an anonymous nuclear det- onation on U.S. soil would be many times greater.49
Third, the strong positive relationship between the number of fatalities stemming from an attack and the rate of attribution (as depicted in agures 1 to 3 above) suggests that the probability of attribution after a nuclear attack— with its enormous casualties—should be even higher. The 97 percent attribu- tion rate for attacks that killed ten or more people on U.S. soil or that of its allies is based on a set of attacks that were pinpricks compared to nuclear ter- rorism. The data in those agures suggest that our conclusions understate the actual likelihood of nuclear attribution.
Fourth, the challenge of attribution after a terrorist nuclear attack should be easier than after a conventional terrorist attack, because the investigation would begin with a highly restricted suspect list. In the case of a conventional terror attack against the United States or an ally, one might begin the investiga- tion at the broadest level with the U.S. Department of State’s list of fifty-one foreign terrorist organizations. In the case of a nuclear terror attack, only afteen of these FTOs have state sponsors—and only one sponsor (Pakistan) has either nuclear weapons or assile materials. (If Iran acquires nuclear weap- ons, that number will grow to two, but there is no overlap between the terror groups that Pakistan supports and those that Iran assists.)
Finally, any operation to detonate a nuclear weapon would involve complex planning and coordination—securing the weapon, learning to use it, planning the time and location of detonation, moving the weapon to the target, and conducting the attack. Even if only a small cadre of operatives knew the nuclear nature of the attack, the planning of a spectacular operation would be hard to keep secret.50 For example, six months prior to the September 11 attacks, Western in- telligence detected numerous indications that al-Qaida was planning a major at- tack. The intelligence was not speciac enough—or the agencies were not nimble enough—to prevent the operation, but the indicators were “blinking red” for months, directing U.S. attention to al-Qaida as soon as the attacks began.51
[ Page 104 ]
The United States and its allies should be able to deter nuclear-armed states from passing their weapons to terrorists, because a terrorist nuclear strike would not remain anonymous for long and would soon be traced back to the originating state. This conclusion is based on two empirical andings. First, among the relevant past cases of conventional terrorist attacks—those targeting the homelands of powerful states and causing significant casualties—almost all were successfully attributed to the perpetrating terrorist organization. Second, linking the attributed terrorist organization to a state sponsor would not be difacult. Few foreign terrorist organizations have state sponsors; those that do typically have only one; and only one suspected state sponsor of terrorism (Pakistan) has nuclear weapons or sufacient stockpiles of nuclear materials.
Furthermore, potential sponsors of nuclear terror face a wicked dilemma: to maintain distance by passing the weapon to a terrorist group they do not know well or trust, or to maintain control by giving it to a group they have co- operated with repeatedly. The former strategy is mind-bogglingly dangerous; the latter option makes attribution from terror group to sponsor simple.
[ Page 104 ]
Our findings have two important policy implications. First, the fear of nuclear attack by proxy by itself does not justify costly military steps to prevent nuclear proliferation. Nuclear proliferation may pose a variety of other risks, and the appropriate level of U.S. efforts to stop proliferation should depend on the cumulative effect of these risks, but the dangers of a nuclear handoff to ter- rorists have been overstated. For example, Iranian leaders would have to be crazy or suicidal to think that they could give a nuclear weapon to one of their terrorist collaborators and face no repercussions. If leaders were that irrational, the bigger problem would be direct nuclear attack without concern for the retaliatory consequences, not the alleged problem of a nuclear handoff.