A lot of things can and should be automated—but nuclear bombers are not one of them.
Unfortunately, it’s not clear that Moscow agrees. Reports surfaced in July that Russia has begun development of a hypersonic nuclear bomber that can deliver nuclear strikes from outer space. Unnamed officials quoted in the semi-official Russian news organ Pravda say that the bomber will have an unmanned variant. Their statement has not been confirmed, but the idea that Russia would pursue an unmanned nuclear bomber is not new. The commander of Russia’s long-range aviation fleet, Lt. Gen. Anatoly Zhikharev, stated in 2012 that Russia was considering developing a “pilotless” sixth-generation nuclear bomber.
While it’s too soon to know for sure whether or not the new Russian bomber will be unmanned, it’s apparent that Russian military officials have been considering that option for some time. And Russian policymakers have made no public promises that the nuclear mission would only be carried out by a manned version of the bomber.
This development is deeply concerning. Deploying a highly autonomous unmanned nuclear bomber would significantly raise the risk of inadvertent or uncontrolled nuclear war. As the world prepares for war in the robotic age, the United States must take steps to ensure that the nuclear mission remains manned.
An unmanned nuclear bomber? The announcement from July of this year leaves much room for skepticism. One thing that should not be taken lightly, however, is the possibility that future Russian nuclear bombers may come with unmanned variants. The recent statement that Russia’s latest bomber will be capable of being unmanned has not been confirmed, but it would align strongly with influential Russian military strategists’ emphasis in recent years on the need for the Russian military to embrace unmanned and autonomous military systems in order to win future wars.
There are plenty of reasons why Russian military thinkers might consider de-manning their nuclear bomber. Most boil down to one thing: Russian policymakers think that having an unmanned nuclear bomber might one day be useful, if not necessary, to protect their country. Looking at the future air and space operating environments, they foresee the possibility that rapidly-improving enemy air and space defenses will make it impossible for manned aircraft—or small numbers of unmanned aircraft, for that matter—to get in range of their targets. To launch nuclear strikes from air or space, then, they might need to use large swarms of robotic systems capable of autonomously navigating to the target; evading or defeating any US and NATO countermeasures they encounter on the way; and releasing their nuclear payloads against previously designated targets. Given Moscow’s history of automating nuclear strike platforms, this calculus has clear precedent.
The problem with autonomy. Assigning the nuclear mission to highly autonomous, unmanned bombers would create an unprecedented risk of inadvertent or uncontrolled nuclear war between the United States and Russia. No matter their sophistication, autonomous systems can behave unexpectedly for a wide variety of reasons, including system malfunction; unanticipated interaction with the air, space, or cyber environments; or hacking by the enemy.
This creates two types of vulnerabilities when it comes to a nuclear bomber. The first is located early in the kill chain—the series of steps taken to find and ultimately destroy the enemy—at the point where the unmanned system is ordered to begin a nuclear strike mission. Due to any of these unexpected inputs, the unmanned bomber could initiate a nuclear strike mission completely against the will of its earthbound operators.
The second vulnerability is located near the end of the kill chain, where the bomber would launch its ordnance at predesignated targets. A frightening number of unforeseen inputs could cause the unmanned system’s original target coordinates to be scrambled or replaced. This could lead it to launch nuclear weapons at targets that were previously off-limits, like major cities.
Having a pilot onboard would create a “human circuit breaker” that could intercede to manually halt operations if something went awry, such as if orders were received to launch a nuclear strike during peacetime, or to hit civilian centers early on in a limited nuclear war. Soviet colonel Stanislav Petrov played this role in 1983 when the sun’s reflection off of the tops of clouds caused an automated early warning system to falsely report that the United States had launched a nuclear attack on Russia. Without a pilot, the bomber would be quite vulnerable to such manipulation.
It may be possible to design automated fail-safes for these systems. But automated fail-safes would be vulnerable to the same types of failures due to technical malfunction, environmental triggers, or hacking. And, depending on when and how the aircraft’s fail-safe engaged—for instance, if it kicked in after the bomber had already begun its final approach on to a target—it may be too late to prevent the defender from initiating its own retaliation sequence.
The potential ramifications of such unexpected behavior would be quite severe indeed. An unauthorized nuclear first-strike by an unmanned bomber would almost certainly trigger retaliation, rapidly forcing the United States and Russia down a path towards nuclear war. A similar effect would occur if a limited nuclear war were ongoing and an unmanned system struck a site beyond the designated set of targets, leading to unintentional escalation. And the potential for a third party to hijack an unmanned bomber in order to trigger nuclear war between the United States and Russia is increasingly real, particularly as advanced cyber capabilities become available to a greater number of state and non-state actors.
Keep the nuclear mission manned. Russian readers might receive this criticism with indignation and point out that the United States hasn’t firmly rejected the possibility of de-manning the nuclear mission either—and they’d be right. Indeed, the US Air Force has offered only ambiguous language on this point, suggesting that it is keeping its options open to de-man nuclear strike assets in the future. This is particularly concerning, given reports that the US Air Force is considering designing an unmanned variant of its own nuclear bomber—the Long-Range Strike Bomber—in the coming years.
But the United States is not immune to the same vulnerabilities that would imperil a Russian unmanned nuclear bomber. For example, it remains unknown why the US RQ-170 stealth drone went down in Iran in 2011. But there have been claims that it was brought down by enemy hacking —something that could not have happened with a pilot onboard. What if the RQ-170 had been a US unmanned nuclear bomber on patrol?
The truth is that no state is immune to the vulnerabilities inherent to autonomous systems—vulnerabilities that would dramatically undermine the reliability of the bomber leg of the nuclear triad. Now is the time to avoid needless catastrophe and set the precedent that the nuclear mission must remain manned during the robotic age.
The United States should clearly and unequivocally reject the possibility of using unmanned nuclear strike assets. It should state forcefully that no potential operational benefits afforded by an unmanned nuclear bomber could outweigh the potential costs of a nuclear conflict driven by the unexpected behavior of a highly autonomous unmanned system. The Defense Department should focus instead on developing ways to penetrate enemy air defenses using manned nuclear bombers, perhaps escorted by highly autonomous, unmanned wingmen.
Washington should then engage Moscow directly on these points. Its objective should be to secure Russia’s entry into an international agreement banning the automation of nuclear strike assets. The agreement should be premised on a mutual understanding of the risks of automating the nuclear mission and confidence-building measures assuring each state of the other’s continued adherence to the agreement.
And the United States should not stop there. With Russia and other partners’ support, the United States should lead arms control negotiations with China, India, Pakistan, and other nuclear-armed states to craft an international agreement prohibiting the de-manning of the nuclear mission. These talks will undoubtedly face significant hurdles, not the least when it comes to defining “autonomy.” But with the automation of nuclear strike assets just over the technical horizon, these discussions must begin now. If the world’s nuclear-armed states wait for real-world events to demonstrate the folly of integrating greater autonomy into their nuclear strike assets, it may be too late.