Key Takeaways
- DoD Directive 3000.09 is the U.S. military’s policy for autonomous weapons which find and attack targets without a human controlling them.
- The policy does not ban these weapons, but establishes rules to make sure they are safe and used correctly.
- The main goal of the Directive is to update the military’s technology while making sure humans stay involved in decisions regarding deadly force.
- Supporters believe the Directive balances innovation with human oversight, while critics argue that vague rules and lack of outside review allows machines to make deadly decisions without accountability.
What is DoD Directive 3000.09?
DoD Directive 3000.09 is the U.S. Department of War’s (formerly the Department of Defense) policy on developing and using autonomous weapons. Autonomous weapon systems (AWSs) are systems that, once activated, can select and engage targets without further human intervention.
The Directive was first written in 2012, making the United States the first country to create detailed rules about AWSs; it was later updated in 2023 to match new AI technologies, clarify which weapon systems need senior-level approval before development, and align with the DoD’s 2020 AI Ethical Principles. The goal of the Directive is to create guidelines that allow the military to develop new autonomous technology while reducing the risk of unintended conflicts.
While the Directive does not ban AWSs, it does create a review process and safety requirements to ensure safe usage.
How does the United States Use Autonomous Weapons Today?
The United States currently uses semi-autonomous and human-supervised systems like the Phalanx CIWS (a defense gun on Navy Ships) and the Centurion C-RAM (protects military bases from missiles). These systems are used for defense in situations where threats appear too quickly for humans to react. The United States doesn’t currently have AWSs that can find and attack targets without human supervision.
Key Provisions of Directive 3000.09
The Directive establishes strict guidelines to ensure AWSs remain under meaningful human control and comply with international law. The Directive includes five core requirements:
- Human Judgement Requirement: All AWSs must be designed so that commanders and operators can use “appropriate levels of human judgement” when deciding whether to engage.
- Testing Requirements: Weapons must go through rigorous testing in realistic combat conditions to make sure they correctly follow the operator’s instructions.
- Senior Level Review: Top military leaders must review and approve AWSs before they’re built and used, with some exemptions.
- Training and Controls: Operators must get proper training and the weapon controls must be easy to understand so operators can make knowing and intelligent decisions.
- Following the Law: All use of AWSs must follow the law of war, international treaties, safety rules, rules of engagement, and the DoD’s AI ethics rules.
How is AI used in Autonomous Weapons?
AI programs are trained through thousands of examples. The weapon is programmed to recognize a target profile, such as the appearance of a military vehicle or how one moves, and sensors look for things in the environment that match this profile. When sensors find something that matches the target profile, the system can attack it; how much human control there is depends on the weapon.
What Do Supporters Say?
Supporters of the Directive argue that it strikes the right balance between innovation and safety. They argue that banning AWSs too early would be a strategic mistake, as it would not allow the United States to compete with other countries actively developing AWSs.
Supporters also point out that AWSs are particularly useful in defensive situations. For example, when incoming missiles travel at extremely high speeds, human operators may not react fast enough to protect ships or military bases. In these cases, AWSs can save lives by responding in milliseconds.
Finally, supporters emphasize that the Directive creates accountability. The Directive requires the approval of senior military leaders before deployment, ensuring that ethical, legal, and safety considerations are formally assessed.
What do Critics Say?
Critics note that the phrase “appropriate human judgement” is too vague. As a result, different operators use different standards when deciding whether to use force. This lack of clarity makes it difficult to hold anyone accountable when things go wrong.
Critics also worry about the lack of external oversight. The policy only requires reviews by people within the military. While there is a requirement to report yearly to Congress, there is no such requirement to inform the public or allow independent experts to examine how these weapons are being developed and used. They worry that this lack of transparency makes it difficult for anyone outside the Pentagon to understand what weapons the United States has, how they work, or what protections are in place. Without transparency, critics argue that there’s no way to know if the rules are being followed.
Finally, critics raise ethical questions about whether machines should ever make life-or-death decisions. Even if the technology works perfectly, they argue there’s something immoral about letting an algorithm decide who lives and who dies. Humans have moral responsibility and can understand context in ways that machines cannot. A soldier might recognize signs of surrender or differentiate a civilian from a threat; autonomous weapons, no matter how advanced, lack this kind of moral judgment and human understanding.
Conclusion
DoD Directive 3000.09 tries to balance military innovation with responsible protections and ethical rules. Today, the debate continues on whether these guidelines provide enough accountability or if the DoD should have clear, strict rules as AWSs become more advanced.
Frequently Asked Questions
Does the United States have fully autonomous weapons that can kill without human involvement?
No. The United States only uses semi-autonomous and human-supervised systems. The United States does not currently deploy fully autonomous lethal systems that independently select and attack targets without human supervision.
What does “appropriate levels of human judgement” actually mean?
The Directive doesn’t clearly define what “appropriate” means, which allows commanders to interpret it differently. For example, one commander might consider it appropriate for an operator to cancel an attack, while another might require direct approval before each decision.
Why doesn’t the Directive ban autonomous weapons completely?
The Directive takes a regulatory approach rather than an outright ban for several reasons. First, countries like China and Russia are actively developing autonomous weapons, and banning them could put the United States at a strategic disadvantage. Additionally, autonomous systems are valuable in defensive situations where human reaction time is too slow. The policy aims to allow continued development while establishing safety requirements to ensure responsible use.
Are other countries developing autonomous weapons, and do they have similar practices?
Yes, many countries are developing autonomous weapons technology, including China, Russia, Israel, and others. Most other countries don’t have publicly available policies as comprehensive as the U.S. Directive, though there are ongoing international discussions about establishing global standards for AWSs.