Military Command Authority in the Age of Artificial Intelligence

The research paper examines how AI-enabled decision-support systems in military operations erode 'command authority'—the ability to bear consequences and exercise independent judgment—even when 'decision authority' (the formal act of authorization) remains with a human. The author contends that AI systems pre-structure the decision space, effectively predetermining outcomes while leaving junior officers to absorb the operational and career risks. This 'compliance trap' is reinforced by institutional incentives that make following machine advice the safest career choice. To counter this 'command drift,' the paper proposes requiring AI systems to declare embedded risk assumptions, providing formal protections for personnel who override machine recommendations, and expanding accountability reviews to audit the advisory layer's influence.

• AI advisory systems preserve the formal act of authorization but erode actual command authority by pre-defining the decision space through training data and model thresholds, meaning the machine 'decided what was decidable' before the human operator intervened.
• There is a structural mismatch where influence over operational outcomes is distributed among upstream actors—such as designers and procurers—while formal accountability remains concentrated with the operator.
• Risk is systematically reallocated 'downward' to junior operators and forward commanders who face the greatest time pressure and least institutional standing, while benefits of efficiency accrue to senior leadership.
• Institutional incentives create a 'compliance trap' where following AI advice is the most defensible career strategy, as overriding a system requires significant justification while compliance requires almost none.
• The 'advisory' designation of AI systems serves as a political tool that allows institutional actors to maintain influence while preserving deniability and insulating themselves from downstream consequences.
• The author proposes three specific reforms to prevent 'command drift': requiring systems to declare embedded risk assumptions, establishing formal protections for those who reject machine recommendations, and auditing how advisory systems shape the conditions of decisions.

APA

The Council on Strategic Risks (2026). Military Command Authority in the Age of Artificial Intelligence. https://councilonstrategicrisks.org/2026/08/27/military-command-authority-in-the-age-of-artificial-intelligence/

Chicago

The Council on Strategic Risks. Military Command Authority in the Age of Artificial Intelligence. 2026. https://councilonstrategicrisks.org/2026/08/27/military-command-authority-in-the-age-of-artificial-intelligence/.

Wikipedia

{{cite report |author=The Council on Strategic Risks |title=Military Command Authority in the Age of Artificial Intelligence |date=27 August 2026 |url=https://councilonstrategicrisks.org/2026/08/27/military-command-authority-in-the-age-of-artificial-intelligence/ |access-date=23 October 2026 |via=Climate Insights Directory}}

BibTeX

@techreport{thecouncilonstrategicrisks2026military, author = {{The Council on Strategic Risks}}, title = {{Military Command Authority in the Age of Artificial Intelligence}}, institution = {The Council on Strategic Risks}, year = {2026}, month = aug, url = {https://councilonstrategicrisks.org/2026/08/27/military-command-authority-in-the-age-of-artificial-intelligence/}, urldate = {2026-10-23}, note = {Indexed by Climate Insights Directory} }