Ethical and Legal Frameworks for Weaponized AI and Robotics

Integrating artificial intelligence into the kill chain represents one of the most significant evolutions in modern warfare. New normative questions are emerging about targeting, including compliance with international humanitarian law, unexamined ethical dimensions, and changes to domestic regulations. The U.S. Department of Defense has seemingly revised its battlefield targeting doctrine to allow artificial intelligence to initiate actions with human oversight, marking a significant shift in how the U.S. military approaches AI-enabled military targeting. 

The new reportedly revised DoD targeting principles envision “systems where AI initiates actions with human monitoring.” This differs from “human in the loop” systems, where a human initiates each action. This policy shift is part of a broader push toward autonomous systems, driven by the speed of future warfare and advances in adversary AI.  

DoD Directive 3000.09 on Autonomy in Weapons Systems was last revised three years ago. President Donald Trump signed National Security Presidential Memorandum 11, titled “Artificial Intelligence in the National Security Enterprise,” on June 5, 2026. The directive mandates the U.S. military and intelligence community to accelerate AI adoption, reversing multiple Biden-era oversight requirements. NSPM-11 requires the Pentagon to update 3000.09 — within 90 days — and institute annual reviews “to account for the rapidly evolving capabilities of AI systems.” The session will examine these policy developments and their potential impact.   

The intersection of these policies governing battlefield AI and technological developments compels a rethinking of empirical and normative components of a targeting decision. Empirical decisions driven by measurable data, technological capabilities, and tactical optimization are part of the equation. But so do normative elements. Key areas to examine during this session include how ethical and legal constraints shape targeting decisions, and how humans have made those decisions in the past. Does our growing dependence on AI for the empirical processing speed preclude human decision-making in the normative realm?   

The emergence of Large Language Models (LLMs) has revolutionized digital transformation, positioning Artificial Intelligence (AI) at its core. LLMs’ exceptional capabilities have enabled diverse applications across industries. However, the ethical and legal examination of AI’s role on the battlefield has not kept pace with evolving battlefield norms. 

This session examines the transformative potential of LLMs in modern warfare and evaluates their applicability in military contexts. It delves into the technical aspects of LLMs, emphasizing their capability to autonomously generate contextually relevant text using deep learning principles. It also discusses potential military applications, such as enhancing decision-making and improving battlefield functions, including intelligence, surveillance, target acquisition, and reconnaissance (ISTAR) integration. 

The potential use of LLMs to assist humans and enhance military processes holds great promise and could offer significant advantages for achieving operational and even strategic objectives. However, bias in training data, accountability for model outputs, and potential hallucinations highlight the importance of maintaining human oversight and responsibility in decision-making. As operations speed up and intensify, it remains unclear whether human operators can stay in control. When mistakes happen, the public is willing to blame LLMs, but whether that is warranted remains an open question. Analyzing incidents, such as the bombing of the Shajareh Tayyebeh girls’ elementary school in Minab, Iran, is key to determining lessons learned whenever tragedy results from LLMs, human error, or both.  

In 2025, a CERL report stressed the importance of DoD programs to mitigate civilian harm amid efforts to slash it.  As a report from the Center for Naval Analysis highlights, countries worldwide have taken early steps to leverage artificial intelligence (AI) in military capabilities and to prevent and mitigate civilian harm. A main concern is whether military AI can distinguish between valid military targets and civilians. 

This session will consider how AI in warfare can strengthen policies to protect civilians and mitigate civilian harm. In 2016, the NATO Policy for the Protection of Civilians (2016) set out that all feasible measures must be taken to avoid, minimize, and mitigate harm to civilians. In 2022, the U.S. Department of Defense introduced the Civilian Harm Mitigation Action Response Plan (CHMR-AP). Yet many questions remain as experts consider how to implement this framework amid the emergence of battlefield AI. The session examines whether AI systems provide the capabilities combat commanders need to mitigate civilian harm successfully.  

This public keynote panel will address the ethical and legal dilemmas of autonomous weapons and artificial intelligence (AI) with special focus on AI-enhanced targeting.  It will examine key questions: What are the promises and perils of emerging technology on the battlefield? How can the United States and other actors ensure AI is used responsibly and in compliance with existing normative and legal frameworks? What kinds of ethical principles should guide the development and deployment of AI and autonomous weaponry? What are the socio-political forces that force governments to examine the inherent tension within emerging technologies and moral and legal choices?  

Unlike semi-autonomous weapons systems, which leave target selection to humans and reserve computer-activated agency for implementing pre-selected aims, true Lethal Autonomous Weapon Systems (LAWS) enable AI-based systems to make targeting decisions. Human judgment often integrates ethical and legal decision-making, but there is no reason to assume AI-based systems can do the same.  

Can LAWS be designed to simulate human ethical decision-making? This question may require reflection on what human beings are actually doing when they engage in ethical reasoning. Given that human ethical decision-making is frequently flawed, the question arises whether AI-based systems should be designed after human ethical decision-making or whether a different modality is preferable for meeting human goals.  Are our goals better met through the human mode of reasoning, which is case-by-case, impressionistic, and based on intuition?  Or might human goals be better served through the use of the top-down, systematic reasoning of machine decision-making, especially in view of the flawed and inconsistent nature of ethical decision-making? The session will examine these questions and how decision speed affects and alters the ability to exercise moral judgment.  

The use of AI technologies in weapons systems has triggered a debate, especially regarding human control, responsibility, and accountability in autonomous and intelligent systems. However, most ethical and legal discussions focus primarily on the point of use of hypothetical AWS.  

Ethical and Legal responsibility and accountability for AWS span the entire life cycle, requiring human oversight from design to decommissioning. Ethical and legal obligations shift across phases, ensuring that military operators, commanders, and engineers remain responsible for and in control of the decisions machines make.  This includes decisions made during design and development, assessments during testing and evaluation, and ultimately decisions during deployment and command.  

The session will examine the use of LAWS and the need to create an effective legal system. This could include a strict liability system, so AI manufacturers must internalize the costs of the externalities associated with their weapons systems. It will also assess the burden of both institutional review and individual accountability that the law of armed conflict imposes on military organizations and the people who use AI-enabled systems. Additionally, the session will examine whether societies and their constituent organizations and citizens have a moral obligation to develop and use autonomous weapons if their adoption can better protect noncombatants from harm in war.