Pentagon Updates Doctrine to Move Toward AI-Initiated Targeting

The updated policy marks a departure from the long-standing "human in the loop" approach, envisioning systems where AI initiates tasks with human monitoring. While Pentagon officials maintain that commanders remain central to every engagement, the document explicitly acknowledges that the speed of modern warfare may eventually demand fully autonomous systems to maintain an advantage against adversaries. This shift aims to compress the "sensor-to-shooter" cycle, leveraging algorithms to process intelligence at a scale far beyond human capacity.

The revised guidelines arrive amid heightened scrutiny over civilian harm, following an investigation into a strike in Minab, Iran, that killed 120 children. Though the doctrine emphasizes the need for ethical frameworks to mitigate risks, it does not currently provide them. Instead, it places the burden of responsibility on commanders to oversee automated outputs. This tension between operational speed and moral accountability remains a point of friction; even partners like Anthropic PBC have cautioned that current technology lacks the reliability required for fully autonomous lethal force. As the military pushes to integrate these tools, the policy acknowledges that automation is an augment to human judgment, not a replacement for the legal and ethical obligations inherent in the law of armed conflict.