<p class="wp-block-paragraph">President Donald Trump instructed the Pentagon to update the rule governing autonomy in weapon systems within a 90-day timeframe, which is set to conclude this week. The directive emphasizes not only the faster adoption of artificial intelligence but also the preservation of the chain of command and operational authority.</p>
<p class="wp-block-paragraph">This directive presents the Pentagon with an opportunity to address a more complex issue than mere definitions of autonomous weapons.</p>
<p class="wp-block-paragraph">One proposal from the Center for Strategic and International Studies suggests expanding the definition of autonomous weapons to include not just drones and munitions but also the software that integrates sensor data, selects targets, and manages the kill chain. This broader definition is seen as necessary since software plays a critical role in weapon autonomy.</p>
<p class="wp-block-paragraph">However, a broader definition is just the initial step. It informs the military about what falls under regulatory guidelines but does not clarify the extent of human authority required once a system is classified as autonomous.</p>
<p class="wp-block-paragraph">Current Directive 3000.09 mandates that autonomous and semi-autonomous weapon systems must allow commanders and operators to exercise appropriate levels of human judgment regarding the use of force. It also requires thorough testing, transparency in system status, and mechanisms for terminating engagements or seeking additional operator input when necessary.</p>
<p class="wp-block-paragraph">The update can enhance these foundations in three key ways.</p>
<p class="wp-block-paragraph">First, it should clearly specify human authority rather than just the degree of human involvement. For each system and mission, the rules should delineate which decisions are reserved for human operators, under what conditions the system can act independently, and which boundaries must not be crossed autonomously. This can be termed an authority envelope, defining what actions are permitted for machines and what decisions remain human responsibilities.</p>
<p class="wp-block-paragraph">Second, these boundaries should be enforced prior to action. If a decision is designated for a human, the system should be technically incapable of crossing that boundary without the necessary authorization. Discovering a safeguard only during an after-action review does not constitute an effective safeguard.</p>
<p class="wp-block-paragraph">Third, human control should be verifiable rather than merely asserted. Systems should maintain sufficient records of decisions to demonstrate the information presented, the machine's recommendations, the human's authorizations, the operational limits in effect, and the sequence of events. The goal is not to automate legal judgment but to ensure that the human role mandated by policy is practically upheld.</p>
<p class="wp-block-paragraph">This approach does not require the Pentagon to slow down AI adoption; it builds on principles already recognized by the department. The AI ethics framework of the War Department calls for systems to be traceable, reliable, and governable, including the ability to disengage or deactivate systems that behave unexpectedly. The new autonomy rule can translate these principles into measurable requirements for systems that impact the use of force.</p>
<p class="wp-block-paragraph">This is significant because the speed of machines alters the meaning of “human in the loop.” A person may be formally present while the decision-making window narrows too much to evaluate evidence, challenge recommendations, or exercise reserved authority before actions are completed. The solution is not to require human input at every moment but to pre-determine which decisions must remain human and to design systems so that these decisions cannot be transferred to software as operational tempo increases.</p>
<p class="wp-block-paragraph">Trump’s memorandum established a clear objective: to accelerate military AI while ensuring accountability and maintaining the chain of command. The Pentagon now has the opportunity to make this objective sustainable beyond any single administration or technology cycle.</p>
<p class="wp-block-paragraph">An effective update to Directive 3000.09 would do more than refine the definition of autonomous weapons. It would clarify for commanders which decisions remain theirs, inform engineers about the necessary boundaries, and guide testers on the evidence needed to demonstrate that human authority is preserved despite machine speed.</p>
<p class="wp-block-paragraph">The goal is not merely to define autonomous weapons but to ensure that human command is effectively maintained.</p>
<p class="wp-block-paragraph"><em>Burak Oktenli is an independent researcher based in Washington, D.C. He holds a bachelor’s degree in computer science and engineering from the University of South Florida and an MBA, and is completing a master’s of professional studies in applied intelligence at Georgetown University. His writing focuses on defense strategy, alliance burden-sharing, military technology, and the governance of autonomous systems.</em></p>