Iran’s proxies are running roughshod over America’s allies and interests in the Middle East. Hezbollah is dictating the terms of Lebanese politics and preparing for war with Israel. In Yemen, Houthis indiscriminately launch missiles into Saudi Arabia. Meanwhile, in areas of Iraq that Iranian-backed militias have liberated from the Islamic State, hundreds of men and boys have disappeared; scores of others have been executed.

But it’s not just Iran that gives proxy war a bad name. In the Cold War, both the United States and Soviet Union employed proxies, often to ill effect. And in the post-9/11 era, proxies continue to play a critical role in how the United States, Russia and Iran seek to realize national-security objectives while trying to limit costs. But dependence on proxies comes with its own risks and moral hazards. Though Dwight Eisenhower once called proxy war “the cheapest insurance you can find,” the last seven decades of American foreign policy reveal otherwise.

At this moment in history—one where, as the recent National Security Strategy indicates, competition between great powers is increasing—evaluating when, where and how the United States decides to support proxies is critical. As great-power competition grows, so does the appeal of proxy war, providing great powers the opportunity to compete while avoiding direct confrontation. But before attempting to beat adversaries at their own proxy game, the United States would benefit from a reckoning with its track record, one that points to many of proxy war’s inherent risks, moral concerns and strategic limitations.

The United States has long operated from the presumption that if a proxy’s cause is just, then America’s support for it must also be just. This ignores a central reality of proxy war: that at the moment a benefactor intervenes, it immediately changes the conflict by changing a proxy’s calculations. With insurance from a powerful benefactor, proxies have greater reason to choose violence over compromise. They have greater reason—and greater resources—to raise a conflict’s stakes, creating risks for themselves and their benefactors.

That’s why proxy war’s first risk to consider is escalation, with America’s role in Vietnam providing a prime example. What began with the commitment of a few dozen military advisors during the 1950s intended to check communist expansion ended two decades later with more than fifty-eight thousand dead Americans and a united, communist Vietnam. Moreover, once a proxy receives additional support and takes greater risks, their opponents are likely to follow suit, calling upon their own benefactors. In fact, in nearly every civil war between 1946 and 2002, no rebel group received weapons from a benefactor without the government they opposed receiving support from benefactors of their own.

The dangers of these escalatory dynamics have featured prominently in debates about how the United States should respond to Russia’s troubling intervention in Ukraine. The Obama administration’s decision to forgo providing lethal aid to forces responding to Russian aggression was deeply unsatisfying, but a potential cycle of escalation and counter escalation weighed heavily in this judgment. So did questions about the endgame, and rightly so.

Consider even the best case: America provides lethal aid that allows pro-Western forces in Ukraine to make rapid gains. Russia would have every incentive to respond, ratcheting up the stakes and costs of the conflict. How far would the United States be willing to go to counter Russia’s increased commitment? In a cycle of escalation and counterescalation, it is difficult to conceive of a situation where the United States would be willing to risk more than Russia over Ukraine. With the Trump administration now reversing its predecessor’s policy, agreeing to send Javelin missiles to pro-American forces in eastern Ukraine, it is critical that the government and the American public fully consider these questions.

This is not to say the risk of escalation should make intervention impermissible—strategically, morally or otherwise. What would be wrong, however, is to be unprepared for it or, in this case, to abandon the Ukrainian government if and when the Russians retaliate in kind. Ideally, such aid should be provided only when steps are taken to ensure their provision leads to greater deterrence, not greater violence. And the question of how much a proxy’s cause actually matters to our own national security—and how far we are willing to go to defend it—should come before the United States provides lethal aid, not after.

The second risk of proxy war comes from diffusion, where weapons, equipment and other support do not stay with the proxy but end up in the wrong hands. Weapons the United States supplied to mujahideen in Afghanistan fighting the Soviets, for instance, played an important role in getting the Soviets out of Afghanistan. They also ended up with adversaries, criminal gangs and terrorist groups as far afield as Bosnia, Iran, Kashmir, Tunisia and Palestine. In fact, only a small portion of the Stinger surface-to-air missiles provided to mujahideen have been recovered. Of course, cooperation with mujahideen proxies also enabled the rise of Al Qaeda, ISIS and other terrorist groups that continue to threaten the United States and its citizens.

In postconflict societies containing proxies that have accepted weapons from a benefactor, these proxies also accept the risk of greater violence within their societies for the long term, with higher rates of murder and other violent crimes in these countries. Again, this is not to say that supplying weapons to a proxy is always the wrong decision. But, whether through better tracking technology or through mechanisms that disable weapons after a certain period of time, these risks can and should be mitigated.

Proxy war’s third and perhaps greatest concern for the United States comes from the problem of dirty hands. For the United States, a nation that draws legitimacy from both the power of its example and its military might, this has special weight.

War is always messy, and proxy war often more so. It can easily make a mockery of our values. Atrocities committed by American-backed right-wing forces in Central America three decades ago, or by America’s partners in the now three-year-old war in Yemen, where a Saudi-led bombing campaign has often failed to discriminate between civilians and combatants, illustrate these risks. If the United States supplies advanced weaponry to proxies without taking necessary steps to prevent their unjust use, our hands are dirty. We too are implicated. And if and when atrocities occur despite our best efforts, a proxy’s feet must be held to the fire. Their leadership has to hold perpetrators accountable.

But there may be cases where the United States has to support a partner that engages in behavior we judge as wrong. Take the campaign to defeat the Islamic State. Here, the United States was well aware that it did not have complete control over the behavior of militias supported by Iran who served as our partners, or for that matter, over the Iraqi army. But allowing vast swathes of territory to remain under the brutality of ISIS was deemed a far greater evil. Of course, this does not allow us to turn a blind eye to injustices committed by those on our own side. And in all proxy wars, the obligation to set a standard of accountability lies with the greater power, the benefactor. Proxies may provide the United States an ability to offload some of its own military costs and responsibilities, but they cannot absolve the nation of its moral responsibilities.

Of course, many will ask why the United States should concern itself with rules when so many threats emanate from adversaries who don’t abide by them, particularly Russia and Iran. In this environment, why should we constrain ourselves with rules and values? Well, it depends on what kind of country we want to be and what kind of world we want to live in.

If we provide advanced weapons to proxies, only for them to create greater instability down the road, how does that serve the nation’s long-term security interests? If we intervene on behalf of a proxy without clarity as to how far we are willing to go, and how important their security is to our own, doesn’t that set us and our partners up for failure? Most critically, if the United States allows proxies to act with impunity, what does the power of our example become?

The appeal of the United States since the Second World War has come from the imperfect pursuit of a noble proposition: that by abiding by certain rules, all nations, large and small, have the ability to grow stronger and more prosperous. Of course, there is a clear alternative: a world in which the strong make the rules. If we support our interests abroad by abandoning the norms that support a just world order, or look the other way as our partners violate them, then we have already lost. We have already accepted the world as our adversaries want to make it.

Dr. C. Anthony Pfaff is the Research Professor for the Military Profession and Ethic at the Strategic Studies Institute, U.S. Army War College. Patrick Granfield, a national-security appointee in the Obama administration, served as a speechwriter for Secretary of State John Kerry and Secretary of Defense Ash Carter. He is currently a lecturer at Georgetown University.