In a paradox worthy of the nuclear age, the conventional capabilities that have enabled the United States to reduce its reliance on nuclear weapons have also increased the risk of misperception that could spark a nuclear war.

The United States now bases its war plans around using its exquisite conventional forces to sever the connections between an adversary’s leadership and its military forces. But in an escalating conflict on the Korean Peninsula, such operations could look to North Korean leader Kim Jong Un like an attempt at regime change — even if the United States did not seek to depose him — and thus induce him to gamble on nuclear use to try to terrify the United States into backing off.

In a war against Russia, meanwhile, the United States might try to protect its satellites by attacking the radars that Russia uses to track space assets. Such radars, however, are also used to detect attacks on Russia’s nuclear forces, thus giving Moscow the ability to launch those forces before they are destroyed. As a result, Moscow might wrongly conclude that attacks on those early-warning radars were the opening salvo of a U.S. campaign to destroy its nuclear forces. Again, a nuclear war could be the result.

Concerns about so-called inadvertent escalation have now reached Congress. As part of the fiscal 2020 National Defense Authorization Act, Congress requested that the undersecretary of defense for policy prepare a report “detailing the Department’s efforts to develop and implement guidance to ensure that the risks of inadvertent escalation to a nuclear war are considered within the decision-making processes with regard to relevant Department activities.”

The Pentagon’s newly released response is titled “Managing Risks of Nuclear Escalation.” Note the (presumably deliberate) absence of the word “inadvertent” in the title. This document is deeply concerned with one particular escalation risk: that the United States’ nuclear-armed adversaries — Russia, China and North Korea — may underestimate U.S. resolve or capabilities. This danger is real, but it has a flipside: that an adversary may assess U.S. war aims to be more ambitious or expansive than they actually are.

Even if one views the United States as the “good guy,” an honest assessment of escalation risks needs to understand the potential perceptions of both sides. The report does not even try to do this.

It rightly notes, for example, the need for U.S. military forces to be survivable, but fails to recognize that stability hinges on mutual vulnerability: that is, in a war between two nuclear-armed states, the risks of an unintentional nuclear escalation are lowest when both sides are convinced that their nuclear forces are highly survivable.

The report’s myopia is again reflected in its final section, which espouses the potential value of dialogues with adversaries, and rightly chides China, Russia and North Korea for various failures to engage. The U.S. interest in increasing “understanding,” however, appears to be monodirectional. Russia, for example, is chastised for seeking “primarily to reduce U.S. power,” yet Moscow’s concerns over the potential for U.S. homeland missile defenses to negate its nuclear deterrent go unaddressed — signaling to Moscow that the United States is uninterested in accepting outcomes short of primacy. The irony here is that even though such defenses are not designed to intercept Russian intercontinental ballistic missiles, and are not capable of doing so reliably, there is scant interest in Washington for even attempting to address Russian concerns.

Going forward, defense planners should reject the idea that the onus falls exclusively on America’s adversaries to take steps to address the risk of inadvertent escalation. The reality is that both the United States and its adversaries may behave in ways that increase the chance of nuclear war’s occurrence. U.S. defense decision-makers have an interest in mitigating this danger by factoring inadvertent escalation risks into decisions about acquisitions, planning, strategic communications, and crisis and conflict management to reduce the risk of U.S. operations generating unintended but highly escalatory threats to an opponent’s regime or nuclear forces.

To be fair, despite the recent report, there are some signs that the Pentagon is growing more aware of the risks of inadvertent escalation. In 2019, for instance, the Defense Department’s report on Chinese military power acknowledged for the first time that “adversary attacks against Chinese conventional missile forces-associated C2 centers could inadvertently degrade Chinese nuclear C2 and generate nuclear use-or-lose pressures.” Notably, the unnamed “adversary” here is the United States.

This observation was not followed by any guidance for American war planning — indeed, policy advice was beyond that report’s purview. But such insights must inform defense planning.

Most importantly, the Pentagon should reflect on how the United States might seek to credibly signal restraint in a crisis. Conveying that the United States does not seek to eliminate China’s or Russia’s nuclear forces or to topple the North Korean regime may mean the difference between de-escalation and unlimited nuclear escalation.

Ankit Panda is a Stanton senior fellow with the Nuclear Policy Program at the Carnegie Endowment for International Peace, where James Acton co-directs the program and holds the Jessica T. Mathews chair.