Autonomy and DoD Directive 3000.09 explained

DoD Directive 3000.09, Autonomy in Weapon Systems, is effective 25 January 2023. It reissues and cancels the 21 November 2012 directive of the same title. The originating component is the Office of the Under Secretary of Defense for Policy. Deputy Secretary of Defense Kathleen H. Hicks approved it. The public copy is cleared for public release on the Directives Division Website. The same PDF is also posted at media.defense.gov.

This article restates what that issuance says. It is not a claim that every DoD AI system is a 3000.09 weapon system.

Status as of 14 October 2026: the cited PDF is still DoD Directive 3000.09, Autonomy in Weapon Systems, effective 25 January 2023. Live automated fetches of esd.whs.mil, media.defense.gov and war.gov returned HTTP 403 this pass; the CRS Defense Primer also 403ed. Claims below are kept from those same official URLs and from the last successful editorial fetch. We found no newer public reissue that replaces the 25 January 2023 directive.

What the directive is for

The purpose statement has three bullets. The directive:

• Establishes policy and assigns responsibilities for developing and using autonomous and semi-autonomous functions in weapon systems, including armed platforms that are remotely operated or operated by onboard personnel.
• Establishes guidelines designed to minimize the probability and consequences of failures in autonomous and semi-autonomous weapon systems that could lead to unintended engagements.
• Establishes the Autonomous Weapon Systems Working Group.

What it applies to — and what it does not

Applicability (Paragraph 1.1.a) covers DoD organizations and the design, development, acquisition, testing, fielding, and employment of autonomous and semi-autonomous weapon systems, including guided munitions that are capable of automated target selection. It also covers the application of lethal or non-lethal, kinetic or non-kinetic, force by autonomous or semi-autonomous weapon systems.

Paragraph 1.1.b lists what this directive does not apply to:

• Autonomous or semi-autonomous cyberspace capabilities
• Unarmed platforms
• Unguided munitions
• Munitions manually guided by the operator
• Mines
• Unexploded explosive ordnance
• Autonomous or semi-autonomous systems that are not weapon systems

That last exclusion is the one this directory needs. A contractor AI tool that handles CUI, a logistics model, or a back-office classifier is not brought under 3000.09 by the text we cite. Deputy Secretary Hicks, in an official press briefing on AI in the Department, said the Department is embedding A.I. in many aspects of our mission, from battle space awareness, cyber and reconnaissance, to logistics, force support, and other back office functions, and separately that our policy for autonomy in weapons systems is clear. Those are two statements, not one scope.

The two definitions that drive the rest

The glossary is for the purpose of this directive:

• An autonomous weapon system is a weapon system that, once activated, can select and engage targets without further intervention by an operator. That includes, but is not limited to, operator-supervised autonomous weapon systems that are designed to allow operators to override operation of the weapon system, but can select and engage targets without further operator input after activation.
• A semi-autonomous weapon system is a weapon system that, once activated, is intended to only engage individual targets or specific target groups that have been selected by an operator. The definition includes systems that use autonomy for acquiring, tracking, and identifying potential targets or providing terminal guidance provided that operator control is retained over the decision to select individual targets and specific target groups for engagement.

An unintended engagement is the use of force against persons or objects that commanders or operators did not intend to be the targets of U.S. military operations.

Human judgment over the use of force

The first policy sentence is the one most often quoted: Autonomous and semi-autonomous weapon systems will be designed to allow commanders and operators to exercise appropriate levels of human judgment over the use of force.

The Congressional Research Service Defense Primer records a US government gloss that “appropriate” is a flexible term and that there is not a fixed, one-size-fits-all level of human judgment. That primer is a CRS product, not the directive. The directive itself does not define “appropriate” as a numeric control threshold.

The same policy paragraph requires rigorous hardware and software verification and validation (V&V) and realistic system developmental and operational test and evaluation (T&E), plus training, doctrine and TTPs, so that systems function as anticipated in realistic operational environments against adaptive adversaries and complete engagements within a timeframe and geographic area … consistent with commander and operator intentions. If they cannot, they will terminate the engagement or obtain additional operator input.

Human-machine interfaces will be readily understandable to trained operators, provide transparent feedback on system status, and provide clear procedures for trained operators to activate and deactivate system functions.

Persons who authorize the use of, direct the use of, or operate these systems will do so with appropriate care and in accordance with the law of war, applicable treaties, weapon system safety rules, and applicable rules of engagement. Deputy Secretary Hicks’s public formulation is There is always a human responsible for the use of force, full stop.

Senior review — who, when, and the exemptions

With the exception of the uses in Paragraphs 1.2.d.(1) through 1.2.d.(4), autonomous weapon systems, including weapon systems with both autonomous and semi-autonomous modes of operation, must be approved by the Under Secretary of Defense for Policy (USD(P)), the Under Secretary of Defense for Research and Engineering (USD(R&E)), and the Vice Chairman of the Joint Chiefs of Staff (VCJCS) before formal development. They must be approved again by the USD(P), the Under Secretary of Defense for Acquisition and Sustainment (USD(A&S)), and the VCJCS before fielding. Those approvals are supplementary to the requirements in other applicable policies and issuances.

The senior review is not required for:

1. Semi-autonomous weapon systems used to apply lethal or non-lethal, kinetic or non-kinetic, force without any modes of operation in which they are intended to function as an autonomous weapon system.
2. Operator-supervised autonomous weapon systems used to select and engage materiel targets for local defense against time-critical or saturation attacks of installations or platforms with personnel.
3. Operator-supervised autonomous weapon systems used to select and engage materiel targets for defending operationally deployed remotely piloted or autonomous vehicles and/or vessels.
4. Autonomous weapon systems used to apply non-lethal, non-kinetic force against materiel targets in accordance with DoDD 3000.03E.

A variant whose system algorithms, intended mission set, intended operational environments, intended target sets, or expected adversarial countermeasures substantially differ from a prior approval will require a new senior review. In cases of urgent military need, listed officials may request a Deputy Secretary of Defense waiver of the senior-review requirements.

The Autonomous Weapon System Working Group supports those reviews. Membership includes USD(A&S), USD(R&E), GC DoD, CDAO, DOT&E and named Joint Staff offices. The Congressional Research Service primer restates the same pre-development and pre-fielding officials.

AI ethical principles sit beside the weapon-system rule

Paragraph 1.2.f says the design, development, deployment, and use of AI capabilities in autonomous and semi-autonomous weapon systems will be consistent with the DoD AI Ethical Principles and the DoD Responsible Artificial Intelligence Strategy and Implementation Pathway. The directive then reprints the five principles as adopted in the February 21, 2020 Secretary of Defense Memorandum. The public announcement of those principles is the 21 February 2020 release. See DoD Responsible AI and Human-machine teaming in DoD and NATO guidance.

CDAO’s public toolkit remains a voluntary process. Completing it is not, on the pages we cite, the 3000.09 senior review.

What official pages do not say

• The directive does not apply to systems that are not weapon systems. We do not treat every defense AI listing on this site as a 3000.09 system.
• The text does not use “human in the loop” as a defined term. It uses appropriate levels of human judgment over the use of force.
• We have not found, on the sources below, a public list of weapon systems that have completed the 2023 senior review. The CRS primer says no weapon system is known to have gone through the senior level review process to date in the edition we fetched. That is CRS, not DoD, and it can go stale.

Questions

Does DoD Directive 3000.09 apply to every DoD AI system?

No. Paragraph 1.1.b says the directive does not apply to autonomous or semi-autonomous systems that are not weapon systems, among other exclusions.

Does “appropriate levels of human judgment” mean a human must approve every engagement?

The directive uses that phrase and requires interfaces that let trained operators activate and deactivate system functions. It does not, in the text we cite, define a single control mode for every system. CRS records a US government statement that “appropriate” is a flexible term.