North Korean leader Kim Jong Un at a ballistic rocket launch drill at an unknown location in an undated photo released by North Korea’s Korean Central News Agency in March 2016. 

There have always been good reasons to worry about nuclear weapons, but those reasons have changed over time. During the Cold War, U.S. national security experts fretted about an expensive nuclear arms race with the Soviet Union. After the 9/11 attacks, specialists and the American public alike were afraid that terrorists might get their hands on highly enriched uranium and make a primitive nuclear device. Those dangers remain. But the first concern has been mitigated to some degree by strategic arms control agreements between the United States and Russia, which are still in place (although not always adhered to). And the second concern has been ameliorated through a significant reduction in the amount of highly enriched uranium used in research reactors around the world. 

Today, however, there is another reason to worry about nuclear weapons: the rise of personalist dictatorships in states that possess or could acquire the bomb. These dictatorships differ from other autocratic governments because their leaders have such dominant personal power that other state institutions—such as parties, politburos, or military officers—cannot overrule the decisions made at the top. Personalist dictators can make decisions on a whim, which creates a grave challenge to the concept of nuclear stability. The world has faced this particular nuclear danger only once before: between 1949 and 1953, when Joseph Stalin enjoyed unchallenged personal dominion over a nuclear-armed Soviet Union. 

Of course, other threats from nuclear proliferation persist. Pakistan’s nuclear arsenal is growing, for example. But it remains in the hands of professional military officers who share at least some degree of power with a democratically elected civilian government. Iran also has latent nuclear capabilities. Yet despite the Trump administration’s unilateral withdrawal from the 2015 agreement that limited Iran’s nuclear activities and the reimposition of U.S. economic sanctions on Iran, the Islamic Republic has, at least for now, decided to keep its commitments to not enrich uranium to bomb-grade levels and to permit international inspectors to monitor any suspected nuclear facilities. 

Personalist dictators are more likely to seek nuclear weapons and, if they get them, more likely than other leaders to use them.

To understand why a nuclear-armed personalist dictatorship poses a much graver danger than those countries, look no further than Kim Jong Un, the eccentric ruler of North Korea. In the six years since he came to power following the death of his father, Kim has solidified his control of the state apparatus and purged potential rivals, including his uncle, whom he executed in 2013, and his half brother, who was murdered in an airport in Malaysia in 2017 by assailants armed with the chemical weapon VX—almost certainly on Kim’s orders. At the same time, Kim has achieved unprecedented success in North Korea’s pursuit of nuclear weapons. After testing a thermo­nuclear device in September 2017 and an intercontinental ballistic missile in November of that year, Kim announced in his 2018 New Year’s address that North Korea had “perfected” its nuclear arsenal and that “the nuclear button is on my office desk all the time.” 

Kim soon entered into direct negotiations with South Korea and, separately, with the United States. Like his three immediate predecessors, U.S. President Donald Trump seeks North Korea’s “complete, verifiable, and irreversible nuclear disarmament.” After meeting with Kim in June, Trump announced that the United States would suspend what he called “tremendously expensive” and “very provocative” military exercises with South Korea and declared that “there is no longer a nuclear threat from North Korea.” In fact, Kim has shown no intention of giving up his weapons, and it is unclear how Washington can achieve its ambitious goal. 

This dynamic is unlikely to remain confined to North Korea. Personalist dictators elsewhere are more likely to seek nuclear weapons in the future and, if they get them, more likely than other leaders to use them. The United States therefore needs to tailor its nuclear doctrine to better deter such leaders—and, if necessary, to fight and defeat them more effectively and ethically. The problem is daunting. The good news is that Washington and its allies have successfully adapted their strategies to meet new nuclear threats in the past, and the steps they must take to do so once again are well within reach. But the bad news is that the Trump administration is not thinking creatively enough and the president is making matters worse by issuing belligerent threats and making unfounded claims of success. 


After 1945, the list of nuclear states grew to include five democracies (the United States, the United Kingdom, France, Israel, and India) and five non­democratic states (the Soviet Union, China, Pakistan, North Korea, and apartheid South Africa). A number of democracies, such as Australia and Sweden, started nuclear weapons programs and then abandoned them, as have a few nondemocracies, such as Brazil and Egypt in the 1970s. Democracies and autocracies alike have joined the 1968 Nuclear Nonproliferation Treaty (NPT), pledging “not to manufacture or otherwise acquire nuclear weapons.”

Yet only autocracies have started or maintained illicit nuclear weapons programs after joining the NPT. These nuclear cheaters were Iran, Iraq, Libya, North Korea, Romania, Syria, Taiwan, and, for a brief period in the 1970s, South Korea. When they began their nuclear weapons programs, all these states were led by autocrats who enjoyed nearly unchallenged authority. Such dictators find nuclear weapons particularly appealing, in part for the usual reason of warding off foreign military intervention, but also because nuclear weapons, unlike conventional ones, provide a way of countering external threats without increasing the risk of internal threats, especially that of a military coup. Such leaders are also less likely to fear the effects of international economic isolation and are not constrained by domestic rivals who might oppose spending scarce resources on a nuclear weapons program. Nor are personalist dictators constrained much by the rule of law, which emboldens them to engage in nuclear cheating, since they face little chance of being outed by internal whistleblowers and because, even if they are caught cheating by foreign powers, they will pay few domestic political costs. 

Yet many of the traits that make personalist dictatorships dangerous also make them incompetent. Such dictators often weaken their state institutions by prizing loyalty over professionalism in military and scientific organizations, thus impeding their nuclear ambitions. In the 1980s, Romania’s laughable nuclear program was run as a pet project by Elena Ceausescu, the wife of the strongman leader Nicolae Ceausescu, who was appointed the head of the National Council for Science and Technology despite having no serious background in scientific research. In Iraq during the late 1970s and early 1980s, the dictator Saddam Hussein executed his deputy prime minister, reportedly for opposing his defense spending plans, and sent a number of senior nuclear scientists to prison because he deemed them insufficiently loyal. During the late 1990s, the Libyan strongman Muammar al-Qaddafi put together a gang that couldn’t proliferate straight: program managers imported the wrong nuclear components because they did not consult scientists first, and no one monitored progress in the program. Indeed, when the International Atomic Energy Agency inspected Libya’s nuclear sites in 2003, they found smuggled-in centrifuges still in their packing crates. 

Only autocracies have started or maintained illicit nuclear weapons programs after joining the NPT.

North Korea’s success therefore represents a watershed. For the first time, a poor and highly personalist dictatorship has developed large numbers of nuclear weapons and long-range missiles to deliver them. North Korea’s persistence, skillful engineering, and extensive support for its scientists helped. So, too, did the illicit assistance that the regime received from the proliferation network run by the Pakistani nuclear physicist A. Q. Khan (which provided centrifuges to enrich uranium) and from companies in Ukraine (which supplied the North Koreans with high-performance liquid-propellant rocket engines). Lastly, Washington failed to get strong global sanctions placed on North Korea until after Pyongyang had already tested its first nuclear weapon, in 2006; by then, it was too late. 

North Korea’s success may now serve as an inspiration. Other governments may calculate that they can copy the North Korean model, especially if Pyongyang offers to carry them across the nuclear threshold, as it has attempted to do at least once in the past. In 2007, the North Koreans were caught helping Bashar al-Assad’s regime construct a secret plutonium-producing reactor in the Syrian desert, which the Israeli Air Force promptly destroyed.


It is difficult to predict which country with a personalist regime—or with a leader who is working to establish such a regime—will be the next to pursue nuclear weapons. Egypt, Syria, and Turkey all seem like contenders. Saudi Arabia might be next in line, if Crown Prince Mohammed bin Salman continues his ruthless consolidation of power when he eventually becomes king. Armed with nuclear weapons, the personalist rulers of these countries will be difficult to deter. They likely value their own lives and those of their family members and cronies more than the lives of their countries’ citizens. They vanquish rivals in order to make their regimes coup-proof and rely on sycophants, often family members, to run their regimes, prizing personal loyalty over professional competence or expertise. A leader surrounded by yes men will have no one who can question faulty assumptions, much less challenge his decision-making authority. 

Recent history demonstrates how such proclivities make personalist dictators particularly likely to mis­calculate. In 1986, Libyan operatives, following Qaddafi’s orders to carry out a campaign of terrorism against Americans, planted a bomb at a nightclub in Berlin popular with U.S. service members, killing two U.S. soldiers and one foreign civilian and injuring 229 other service members and civilians. In response, the United States launched air strikes against military targets in Libya and the compound outside Tripoli where Qaddafi lived with his family. In 1980, Saddam decided to attack Iran without consulting his advisers (resulting in an eight-year-long war), and in 1990, he ordered an attack on Kuwait after consulting with only his son-in-law (leading to the humiliating Persian Gulf War). Saddam even forbade his intelligence agencies from providing reports on the United States, telling them that intelligence was his “specialty.” (He also elaborated on the sources of his unique insight: “some of it out of deduction, some of it through invention and connecting the dots, all without having hard evidence.”)

Atomic autocrat: Kim Jong Un watching a missile launch in Pyongyang, September 2017

Flawed decision-making of this sort also makes personalist regimes accident-prone. According to North Korean government pronouncements, Pyongyang has a preemptive military doctrine, which calls for striking first if Kim receives intelligence that a U.S. attack is deemed imminent and unavoidable. But no outsiders know the exact indicators on which Kim would base his decision. Perhaps he might react to a formal warning issued by trusted organizations within the state. But even in technologically sophisticated societies, these are imperfect. In January, for example, the Hawaii Emergency Management Agency issued a false alarm: “Ballistic missile threat inbound to Hawaii. Seek immediate shelter. This is not a drill.” Throughout the islands, citizens panicked, some running for the beaches, others (more appropriately) sheltering inside their homes. In Washington, fortunately, no one panicked: the U.S. military’s sophisticated sensors did not detect an inbound missile, highly professional military officers quickly reported up the chain of command that the Hawaiian agency had made a mistake, and no high-level official believed that Kim would launch an unprovoked nuclear attack on Hawaii. 

But just imagine what would have taken place had a similar false alarm occurred in Pyongyang rather than Honolulu. North Korea’s missile warning system relies on archaic Soviet radar technology. The North Koreans lack the multiple and independent satellite-based warning systems that create redundancy and reliability for the United States: if someone in North Korea issued an erroneous warning of an attack, no alternative system would correct it. And it’s unlikely that the military in North Korea would report a serious mistake, because if a bureaucrat or a military officer makes an error in North Korea, he doesn’t just get fired; he might also get executed. Finally, Kim is likely to believe that the United States would launch a first strike against North Korea thanks to Trump’s frequent over-the-top threats to do just that.


Even though a nuclear North Korea will remain a dire threat, there are ways to reduce the likelihood of further proliferation. The United States should never rule out using military force against would-be proliferators if they are caught cheating, but diplomacy is always preferable. There is ample room for improvement on the diplomatic front. In 2003, the George W. Bush administration created the Proliferation Security Initiative, through which more than 100 countries coordinate intelligence and interdiction efforts to prevent the smuggling of components for weapons of mass destruction. China, however, is not a member. During the Obama administration, the United States, the Netherlands, and South Korea hosted a series of summits where more than 40 countries with nuclear power facilities shared best practices regarding security, training, and equipment. But Russia dropped out of the process after its invasion of Crimea in 2014. And North Korea’s success in developing nuclear weapons has demonstrated that these efforts were insufficient. That’s why Washington must work with its allies and partners—and also with rivals such as China and Russia—to establish even stricter export controls and counter-smuggling measures.

One improvement would be to require all NPT members to ratify the so-called Additional Protocol of the International Atomic Energy Agency, which mandates that signatories permit inspectors to enter any suspected nuclear facilities on demand. There are many holdouts against universal ratification of the Additional Protocol, including states that may seek nuclear weapons, such as Egypt, Saudi Arabia, and Syria. In 2020, the UN will host the next NPT Review Conference, where all member states will meet to discuss how to improve nonproliferation. In the run-up to the conference, Washington, Beijing, and Moscow should agree to make a major push for universal ratification of the Additional Protocol and should pursue a coordinated bargaining strategy, offering nonnuclear states improved access to nuclear technology in exchange for agreeing to inspections. 

In addition, the United States should develop a common strategy with China, Russia, and other nuclear technology exporters to ensure that countries constructing civilian nuclear power plants for the first time abstain from also taking steps consistent with pursuing a nuclear weapons program—namely, enriching uranium or reprocessing plutonium. This will be a severe challenge since the major exporters—China, France, Russia, South Korea, and the United States—have a clear incentive to prioritize sales over security and to not impose rules on importers. At a minimum, Washington should refuse to help countries acquire nuclear energy if they do not sign and ratify the Additional Protocol and agree not to enrich uranium or reprocess plutonium. 

When it comes to the threat of nuclear-armed dictatorships, multilateral diplomacy can do only so much. Washington also needs to update its approach to deterrence and its nuclear arsenal. In an interview with ABC News in August 2017, H. R. McMaster, Trump’s national security adviser at the time, expressed intense skepticism about the possibility of deterring Kim. “Classical deterrence theory, how does that apply to a regime like the regime in North Korea?” he asked. 

A regime that engages in unspeakable brutality against its own people? A regime that poses a continuous threat to its neighbors in the region and now may pose a threat, direct threat, to the United States with weapons of mass destruction? A regime that imprisons and murders anyone who seems to oppose that regime, including members of [Kim’s] own family [by] using [VX] in a public airport?

The answer is that the United States can deter such a regime not by threatening its subjects but by threatening its leader. Washington must make clear that it will respond with military force only to acts of aggression and that it will target only the dictator himself, the regime’s leadership, and its military forces. And it should discourage senior military officers in such personalist dictatorships from following any rash and suicidal orders by offering them “golden parachutes” if they disobey. This will not be easy. The Obama administration’s 2010 Nuclear Posture Review walked that fine line by stating that any country that “uses chemical or biological weapons against the United States or its allies and partners would face the prospect of a devastating conventional military response” and that “any individuals responsible for the attack, whether national leaders or military commanders, would be held fully accountable.” A new U.S. nuclear doctrine should make explicit what the Obama doctrine implied: any military commander in a personalist dictatorship who disobeys a command to use nuclear weapons will not be held responsible for the consequences of his leader’s aggression. 

The Trump administration’s 2018 Nuclear Posture Review also got this aspect of deterrence right, by threatening retaliation against the appropriate target: 

For North Korea, the survival of the Kim regime is paramount. Our deterrence strategy for North Korea makes clear that any North Korean nuclear attack against the United States or its allies and partners is unacceptable and will result in the end of that regime. There is no scenario in which the Kim regime could employ nuclear weapons and survive.

But Trump himself has repeatedly threatened to start a war with North Korea and to harm the North Korean people rather than just their leaders. In August 2017, Trump declared that “North Korea best not make any more threats to the United States. . . . They will be met with fire and fury like the world has never seen.” At the UN in September 2017, he warned that he might “totally destroy North Korea,” which sounded grossly indiscriminate. And in a press conference in May 2018, Trump threatened to start a preventive war against North Korea that would target the entire country: “In Libya, we decimated that country,” he said. “That model would take place [in North Korea] if we don’t make a deal, most likely.”

The Trump administration should call for the broader development of more lower-yield nuclear warheads and advanced conventional weapons.

Despite Trump’s loose talk, his administration’s 2018 Nuclear Posture Review recognized the need to tailor deterrence to specific adversaries. For example, it called for the development of nuclear warheads with smaller yields for U.S. submarines to counter new Russian lower-yield weapons. Some eminent experts, such as William Perry, a former U.S. secretary of defense, and George Shultz, a former U.S. secretary of state, have criticized this step as making nuclear war more likely. But the logic of deterrence suggests the opposite is true. By enabling a limited and smaller-scale U.S. nuclear response, lower-yield weapons would enhance the credibility of a U.S. threat to retaliate and thus make aggression by Russia less likely. 

The Trump administration, however, missed an opportunity by neglecting to call for the broader development of more lower-yield nuclear warheads and advanced conventional weapons to reduce the collateral damage in a future conflict with a proliferator. Such weapons would make U.S. deterrence both more ethical and more effective—more ethical because they could be used to kill only the leaders and military personnel responsible for acts of aggression, and more effective because they would make the possibility of U.S. retaliation inherently more credible. 


Washington should always prefer conventional military options over nuclear ones. Yet as long as the United States possesses nuclear weapons, it must have war plans for how to use them when necessary. Such plans should always conform to the laws of armed conflict and the just war principles of never deliberately targeting noncombatants, adjusting the use of force in proportion to the threat, and doing everything feasible to spare the lives of innocent civilians. 

In an era of nuclear-armed personalist dictators, the United States should adopt what the arms control expert Jeffrey Lewis and I have termed “the nuclear necessity principle.” Washington should not aim nuclear weapons against any target that could be effectively destroyed with conventional weapons. And if the U.S. military does determine that it needs to attack a target that is so deeply buried or otherwise hardened that it cannot be destroyed with conventional weapons, it should use the lowest-yield nuclear weapon possible to accomplish the mission. 

The historian Alex Wellerstein has developed a website called NUKEMAP that allows users to estimate the civilian fatalities that would result from a nuclear strike anywhere in the world (not including the longer-term deaths that would result from radioactive fallout). NUKEMAP bases its results on data from the 1945 attack in which the United States dropped a 15-kiloton bomb on Hiroshima, instantly killing more than 70,000 civilians. The website lets users adjust the yield of the weapon deployed in any hypothetical strike and thus grimly demonstrates the significance of lower-yield nuclear weapons. NUKEMAP can’t predict whether a nuclear strike would destroy any particular target. Still, the exercise is sobering and revealing. For example, a U.S. attack with a one-megaton bomb on the North Korean command-and-control bunker near the town of Chunghwa, 20 miles from Pyongyang, would kill approximately 37,500 civilians, according to NUKEMAP, whereas a 100-kiloton weapon would immediately kill some 16,000 civilians. An attack on the same site with a 6.5-kiloton weapon (the reported size of a new warhead that the Trump administration has proposed building) would kill around 2,900 civilians—still a terrible toll, but far lower. Using a one-megaton bomb to destroy the tunnels near Sunchon, a city 35 miles north of Pyongyang where the North Koreans have test-launched long-range missiles, would produce about 70,000 immediate civilian deaths. A 100-kiloton warhead would cause 5,700 fatalities. A 6.5-kiloton warhead would immediately kill approximately 800 civilians—again, a dreadful outcome, but far less tragic.

In addition to making U.S. deterrence more ethical, a more discriminate doctrine and the development of lower-yield weapons would allow Washington to better assure its allies that it is neither too cautious nor reckless. Developing lower-yield weapons and more clearly articulating limits on their use would also demonstrate Washington’s commitment under the NPT to work in good faith toward the eventual elimination of nuclear weapons. (The Trump administration’s 2018 Nuclear Posture Review conspicuously failed to mention that pledge.)

A more ethical targeting doctrine would also reduce the risk that a personalist dictator might think that the United States could be “self-deterred” by concerns about civilian deaths. In reality, the American public would likely demand vengeance if the United States or its allies were attacked even in a limited way. And Americans are quite hawkish on the use of nuclear weapons: surveys that the political scientist Benjamin Valentino and I commissioned in 2015 found that nearly 60 percent of Americans would approve of a nuclear strike against an Iranian city that would kill two million civilians in order to avoid a land war that might kill up to 20,000 American soldiers. 

The final reason to support this change in U.S. nuclear doctrine is because it is the right thing to do. The arc of history should be bent, slowly but surely, toward just war doctrine.


After the Cold War, many politicians and scholars thought that the danger of nuclear war had receded. In the years that followed, a number of states went nuclear (or tried to), but the threat of nuclear war seemed to remain far lower than it had been in the decades after World War II. In a 2009 speech in Prague, U.S. President Barack Obama renewed the United States’ commitment to work toward “a world without nuclear weapons.” It was a brave and lofty vision.

What a difference a decade makes. Today, thanks to North Korea’s breakthrough, the world faces a future in which unpredictable, unconstrained personalist dictators might hold the fate of millions of people in their hands. The United States should remain committed to the distant goal of disarmament. But in the meantime, Washington will have to be much smarter about tailoring its nuclear arsenal and its nuclear doctrine to meet this current challenge.

You are reading a free article.

Subscribe to Foreign Affairs to get unlimited access.

  • Paywall-free reading of new articles and a century of archives
  • Unlock access to iOS/Android apps to save editions for offline reading
  • Six issues a year in print, online, and audio editions
Subscribe Now
  • SCOTT D. SAGAN is Caroline S. G. Munro Professor of Political Science, Mimi and Peter Haas University Fellow in Undergraduate Education, and a Senior Fellow at the Center for International Security and Cooperation and the Freeman Spogli Institute at Stanford University.
  • More By Scott D. Sagan