Overview of US Department of Defense Directive 3000.09
The US Department of Defense Directive 3000.09, titled Autonomy in Weapon Systems, establishes the policy framework for the development, testing, and deployment of autonomous and semi-autonomous weapon systems within the US military. Issued initially in November 2012 and updated on January 25, 2023, the directive reflects evolving technological capabilities and ethical considerations surrounding AI-driven weaponry. Its core principle mandates that human judgment must retain authority over the use of force in all weapon systems, aligning with broader Department of Defense AI ethics principles.
The directive applies specifically to weapon systems that incorporate autonomy, whether fully autonomous (human out of the loop) or semi-autonomous (human in the loop or on the loop). It excludes non-lethal systems, unarmed platforms, unguided munitions, manually guided munitions, mines, unexploded ordnance, and non-weapon autonomous systems. The 2023 update emphasizes the integration of AI ethics, safety, and accountability in weapon system design and operational use.
Definitions and Categorization of Autonomous Weapon Systems
DODD 3000.09 defines autonomous weapon systems based on the degree of human involvement in critical decisions such as target selection and engagement.
- Lethal Autonomous Weapon Systems (LAWS): These are weapon systems that, once activated, can independently select and engage targets without further human intervention. This category is also referred to as human out of the loop or full autonomy. The directive explicitly requires that such systems preserve human authority over the use of force, ensuring compliance with laws of war and rules of engagement.
- Human-Supervised Autonomous Weapon Systems: These systems operate with human oversight, allowing operators to monitor and intervene in target engagement decisions. This is known as human on the loop.
- Semi-Autonomous Weapon Systems: These systems require human operators to select targets or target groups before the weapon system engages autonomously. Examples include fire-and-forget missiles, which autonomously deliver effects to pre-identified targets. The directive clarifies that semi-autonomous systems must still adhere to human judgment in determining the appropriateness of engagement.
The directive does not impose restrictions on autonomous or semi-autonomous systems that do not involve lethal or kinetic force, such as cyber capabilities or unarmed platforms. This distinction underscores the focus on preserving human control over decisions that directly result in harm or destruction.
Role of Human Judgment and Operational Safeguards
A central tenet of DODD 3000.09 is the requirement that all autonomous and semi-autonomous weapon systems be designed to allow commanders and operators to exercise appropriate levels of human judgment over the use of force. The term appropriate is intentionally flexible, recognizing that the degree of human involvement may vary depending on the weapon system, operational context, domain of warfare, and specific functions within the system.
The directive specifies that human judgment does not necessarily equate to manual control of the weapon system. Instead, it emphasizes broader human involvement in determining the how, when, where, and why of weapon employment. This includes ensuring that engagements comply with the law of war, applicable treaties, weapon safety rules, and rules of engagement.
To support informed decision-making, DODD 3000.09 mandates that adequate training, tactics, techniques, procedures, and doctrine be developed and periodically reviewed. These resources must enable operators and commanders to understand the functioning, capabilities, and limitations of autonomous systems in realistic operational conditions.
Weapons Review and Senior-Level Approval Process
The directive establishes a rigorous review process for autonomous and semi-autonomous weapon systems to ensure their safety, reliability, and compliance with ethical and legal standards. Before development and fielding, these systems must undergo testing and evaluation to verify that they:
- Function as intended in realistic operational environments, including against adaptive adversaries employing countermeasures.
- If a system fails to meet these criteria, it must terminate the engagement or seek additional human input before proceeding.
- Demonstrate sufficient robustness to minimize the probability and consequences of failures, particularly in dynamic or unpredictable conditions.
- Retain safety features and operational integrity after any changes to their operating state, such as updates or the integration of machine learning capabilities.
Beyond standard weapons reviews, DODD 3000.09 introduces a senior-level review process for autonomous and semi-autonomous systems.
- The Under Secretary of Defense for Policy (USD(P)),
- The Vice Chairman of the Joint Chiefs of Staff (VCJCS), and
- The Under Secretary of Defense for Research and Engineering (USD(R&E)).
These officials must approve the system before formal development begins. Prior to fielding, additional approval is required from USD(P), VCJCS, and the Under Secretary of Defense for Acquisition and Sustainment (USD(A&S)).
The directive also establishes the Autonomous Weapon System Working Group, comprising representatives from USD(P), USD(R&E), USD(A&S), the DOD General Counsel, the Chief Digital and AI Officer, the Director of Operational Test and Evaluation, and the Chairman of the Joint Chiefs of Staff. This group supports and advises the senior-level review process, ensuring comprehensive oversight of autonomous weapon system development.
In cases of urgent military need, the Deputy Secretary of Defense may waive the senior-level review process. However, such waivers are subject to congressional notification requirements, as outlined in the FY2026 National Defense Authorization Act (NDAA).
AI Safety and Compliance with Ethical Principles
The 2023 update to DODD 3000.09 introduces explicit requirements for AI safety in autonomous weapon systems. The directive now mandates that AI capabilities in these systems be designed and engineered to:
- Fulfill their intended functions while detecting and avoiding unintended consequences.
- Possess the ability to disengage or deactivate if unintended behavior is detected during deployment.
This emphasis on AI safety aligns with the broader DoD AI Ethical Principles, which prioritize responsibility, equity, traceability, reliability, and governance in AI system development.
Congressional Notification and Reporting Requirements
The FY2024, FY2025, and FY2026 National Defense Authorization Acts (NDAAs) impose additional transparency and accountability measures on the implementation of DODD 3000.09.
- Section 251 (FY2024 NDAA): Requires the Secretary of Defense to notify congressional defense committees within 30 days of any changes to the directive. The notification must include a description of the modification and the rationale behind it.
- Section 1066 (FY2025 NDAA): Mandates an annual report to congressional defense committees, detailing the approval and deployment of lethal autonomous weapon systems by the United States. This reporting requirement remains in effect through December 31, 2029.
- Section 1061 (FY2026 NDAA): Amends the US Code to require congressional notification of any waiver issued under DODD 3000.09, further enhancing legislative oversight of autonomous weapon system policies.
Required actions
- Autonomous and semi-autonomous weapon systems must allow commanders and operators to exercise appropriate levels of human judgement over the use of force - Senior-level review and approval before formal development and again before fielding; rigorous verification, validation and testing - Design against unintended engagements and adversary interference; comply with the DoD AI Ethical Principles