"I don’t see how an effective arms control regime is possible," says Frank Kendall, former Air Force Secretary, in an excerpt from his book.
The definitional problem Kendall lays out
Kendall opens from a humanitarian and human-rights viewpoint and frames the debate around one inescapable technical fact: lethal autonomy resists clean definition. He argues that trying to draw a line between acceptable and banned systems fails because core capabilities—target detection, intent inference, and engagement planning—exist on continuums with no sharp boundaries. Defensive and offensive labels blur when an autonomous air-defense system can engage a civilian airliner, he notes, and degrees of autonomy lack an obvious, enforceable cutoff.
Why enforcement would be nearly impossible
Even if negotiators could agree on definitions, Kendall says enforcement would be cripplingly difficult. The practical distinction between human control and full autonomy is a decision mechanism—a trigger—that in modern systems will likely be digital and remotely located. That makes covert substitution trivial: "somewhere in the weapon system include the capacity to close that switch autonomously," he writes. Preventing covert changes, he argues, would require "extremely detailed design information and extremely intrusive inspection regimes" that nations are unlikely to accept and would be next to impossible to implement.
Accountability as an alternative to a ban
Rather than a prohibition, Kendall proposes strengthening accountability. He urges holding responsible people and countries liable for violations of the law of armed conflict regardless of the tools used. That list of potentially accountable actors is comprehensive in his formulation: designers, testers, acquirers, fielders, military operators, and political decision-makers all have a duty to ensure compliance. He suggests several concrete legal and administrative measures: raising the bar for acceptable collateral damage, applying negligence or intent standards familiar from criminal and personal-injury law, implementing strict liability in some cases, and designating specific officials and organizations responsible for testing and verifying compliance—analogous to airworthiness certification.
Technological momentum and battlefield realities
Kendall is blunt about momentum. He writes that most enabling technologies for lethal autonomy will proceed "for independent reasons" and that U.S. military service efforts and DARPA programs are advancing capabilities that push toward lethal autonomy. He recounts that representatives of "several liberal democracies" have publicly stated they will pursue relevant technologies out of fear that less principled nations will do so. He also points to open-literature accounts of new unmanned lethal systems and improvements in pattern (target) recognition as daily evidence that "this genie will not be put back in the bottle."
Tactical implications: expendable unmanned systems
Accepting lethal autonomy, Kendall argues, opens doctrinal options unavailable to manned systems. He highlights the permissibility of sacrificing unmanned systems for tactical advantage when cost is not prohibitive. He invokes a conversation with retired Army Major General Bob Scales about exposing the enemy to attack and notes his own Pentagon convening of general officers to explore replacing dangerous infantry tasks with relatively inexpensive, expendable unmanned systems. Kendall points to the Ukraine conflict as demonstrating that such uses "isn’t just attractive, it’s required."
What this means for policymakers, military operators, humanitarians, and adversary states
- Policymakers and regulators: Expect pressure to design legal frameworks that focus on liability, testing protocols, and higher standards for acceptable collateral effects, including proposals for designated certification authorities analogous to airworthiness.
- Military operators, acquirers, and designers: Kendall places responsibility squarely on those who design, test, field, and operate systems to ensure compliance with existing law of armed conflict standards—and suggests the possible imposition of strict liability in some cases.
- Humanitarians and the legal community: Rather than seeking an arms-control ban, Kendall urges refining legal standards and enforcement mechanisms to hold people accountable for indiscriminate autonomous killings, a path that will demand significant legal theorizing about chain-of-control liability.
- Adversary nation-states and groups: Kendall stresses that many nation-states and groups will not hesitate to embrace lethal autonomy and that democratic governments are already pursuing technologies out of concern that others will exploit them.
Kendall’s calculus is stark: banning lethal autonomy is, in his view, both impractical and unenforceable; accountability is possible and preferable; and the technological and operational tides make restraint unlikely. He ends by inviting legal and institutional work to define responsibility and raise operational standards—leaving a pointed question for policymakers and lawyers alike: if the tools cannot be constrained effectively, can the people who wield them be constrained instead?
