UN Talks on Regulating Killer Robots by 2028
States Agree to UN Talks on Regulating “Killer Robots” by 2028
States have agreed to pursue United Nations talks aimed at regulating autonomous weapons systems, commonly known as “killer robots.” The negotiations are targeted for completion by 2028, creating a potential pathway toward international rules for weapons that can perform target-selection or engagement functions with limited direct human control. Source 1
The decision does not create a global ban or a completed treaty. It advances a diplomatic process in which governments must decide which systems should be covered, how much human control is required, and which uses should be prohibited or restricted.
The debate involves military technology, artificial intelligence, international humanitarian law and national security. It also raises a fundamental question: should a machine be allowed to make a decision that directly determines whether a person lives or dies?
What Are “Killer Robots”?
Defining Autonomous Weapons Systems
Autonomous weapons systems can perform certain functions without direct human control at every stage of an operation. Depending on their design, they may use sensors, software and programmed parameters to detect objects, classify potential targets, track movement or engage a target after activation.
Autonomy exists on a spectrum. A weapon may automate navigation while leaving target selection to a human operator. Another system may identify and select targets within a defined area. A third may perform several operational functions with limited human intervention.
This distinguishes autonomous weapons from remotely operated systems. A remotely operated drone is controlled by a person who decides where it moves and what it attacks. An autonomous system may make some of those decisions itself after receiving an initial instruction.
Why the Term “Killer Robots” Is Used
“Killer robots” is an informal term used by campaigners, researchers and media organizations to describe lethal autonomous weapons. It highlights the possibility that machines could make life-and-death decisions.
Official discussions generally use terms such as:
- Autonomous weapons systems
- Lethal autonomous weapons systems
- Weapons systems with autonomous functions
- Autonomous weapons
The public term communicates the central concern quickly, but it can obscure differences between weapons with limited automation and systems capable of independently selecting and attacking human targets.
How These Systems Could Operate
An autonomous weapon could potentially:
- Detect objects, vehicles or people using sensors.
- Classify what it detects according to programmed criteria.
- Track movement across a defined area.
- Select a target within authorized limits.
- Engage the target after activation.
These functions do not necessarily occur in every system. A weapon may have autonomy in navigation but not targeting, or it may use automated target recognition while requiring human authorization before an attack.
The UN talks will need to address these distinctions. A definition that is too narrow could leave important systems outside the rules. A definition that is too broad could cover ordinary military automation that does not make independent lethal decisions.
What States Have Agreed to Do
UN Negotiations Targeted for Completion by 2028
States have agreed to pursue UN negotiations aimed at regulating autonomous weapons, with a target of reaching an agreement by 2028. The reported process concerns weapons that may select and attack targets without direct human control. Source 7
The 2028 date is a negotiating target, not a guaranteed treaty deadline. Governments must still agree on the scope, legal form and obligations of any future instrument.
The outcome could be:
- A legally binding treaty
- Prohibitions and restrictions
- A political declaration
- Voluntary guidelines
- A framework combining binding and non-binding commitments
The Agreement Does Not Yet Create Binding Rules
The decision does not itself:
- Ban all autonomous weapons
- Require states to dismantle existing systems
- Establish universal technical standards
- Create a finalized inspection regime
- Set penalties for violations
- Require immediate changes to military procurement
Those obligations would arise only if states adopt a future instrument containing them. Until then, existing international law and national rules remain central to how governments assess autonomous weapons.
Why the United Nations Is the Main Forum
The UN provides a forum for addressing international security and humanitarian issues with broad participation. It can help governments develop common terminology, compare national policies and negotiate shared limits.
The organization also brings competing perspectives together. Some governments may focus on civilian protection and accountability. Others may emphasize military needs, technological development or the possibility that autonomous systems could improve operational efficiency.
Why Governments Are Debating Autonomous Weapons
Humanitarian Concerns
The central humanitarian concern is the delegation of lethal decisions to software. A machine may process data quickly, but it does not understand context in the same way as a human decision-maker. This raises questions about whether an autonomous system can distinguish civilians from combatants or assess whether an attack would cause excessive civilian harm.
The debate involves key principles of international humanitarian law:
- Distinction between military targets and civilians
- Proportionality in attack
- Precautions during military operations
- Protection of people who are hors de combat
- Accountability for unlawful harm
Autonomous systems are not automatically unlawful. Existing law may apply to their development and use. The unresolved issue is whether current rules provide sufficient protection when software performs functions traditionally handled by human personnel.
Accountability and Responsibility
Autonomous weapons could make it harder to determine responsibility after an unlawful attack. Developers, manufacturers, military personnel, commanders, operators and states may all contribute to the outcome.
A developer may not control battlefield conditions. A commander may not know precisely how a complex system will respond in every situation. An operator may have little ability to intervene after activation.
Future regulation may therefore require a clear chain of responsibility and records that allow investigators to reconstruct a system’s actions, the human decisions surrounding deployment and the conditions present during an incident.
Human Control Over the Use of Force
“Meaningful human control” is likely to remain one of the central concepts in the talks. The phrase suggests that people should retain sufficient control over decisions involving lethal force.
Possible standards include:
- Human approval before every attack
- Human supervision during a mission
- Limits on targets, locations and operating time
- A reliable ability to intervene or deactivate the system
- Human review of instructions and expected behavior
- Restrictions on use in areas containing civilians
A broad political commitment may be easier to negotiate than a precise legal definition. However, vague wording could make the rule difficult to enforce.
Strategic and Security Risks
Governments and experts have raised concerns that autonomous weapons could accelerate military decision-making and reduce the time available for human judgment. Other possible risks include:
- Faster escalation during a crisis
- Deployment of large numbers of relatively low-cost systems
- Pressure to respond before an opponent gains an advantage
- New incentives for an artificial intelligence arms race
- Difficulty controlling systems after activation
- Greater danger from hacking, spoofing or malfunction
These are risks identified in the policy debate, not guaranteed outcomes. Some states may argue that automation could improve defensive capabilities, reduce risks to military personnel or support faster responses to incoming threats.
What the UN Talks May Need to Address
Scope, Prohibited Uses and Restrictions
Negotiators must determine which systems fall within the regulatory framework. Possible categories include fully autonomous lethal weapons, systems with autonomous target-selection functions, defensive systems, uncrewed vehicles, coordinated swarms and existing weapons with automated targeting features.
Rules could focus on technical characteristics, intended use or the level of human control. States might consider:
- A complete ban on particular autonomous weapons
- A prohibition on systems that cannot be meaningfully controlled
- Restrictions on attacks against people
- Limits on use in populated areas
- Rules for specific functions rather than entire platforms
- Restrictions based on target type, environment or duration
The available summaries do not establish which approach states support. The negotiations will determine whether governments seek a preventive ban, targeted restrictions, voluntary controls or a combination of measures. Source 9
Testing, Review and Deployment Controls
A future framework could require testing under realistic conditions, legal reviews before deployment, reliability assessments, cybersecurity safeguards, fail-safe mechanisms, defined operating limits and procedures for system failure.
International humanitarian law already requires states to conduct legal reviews of new weapons in relevant circumstances. New negotiations could clarify how those reviews should address machine learning, adaptive behavior and complex software.
Verification may be difficult because algorithms, source code and military capabilities are often classified. A workable framework may therefore focus on procedures, documentation and national compliance rather than disclosure of sensitive technical information.
Transparency and Accountability
Possible transparency measures include national reports, voluntary disclosures, shared terminology, information exchanges, testing notifications and reports on incidents or near misses.
Post-incident accountability could require preserving system logs, reviewing command decisions, recording human authorization, identifying responsible officials, investigating civilian harm and providing remedies where appropriate.
Rules must identify responsibility across the weapon’s full lifecycle: design, testing, authorization, deployment and use.
Why Reaching an Agreement by 2028 Could Be Difficult
States have different national security priorities. Some may support a preventive ban, while others may prefer voluntary guidelines or argue that existing international law is sufficient. Military powers may resist restrictions that they view as strategically disadvantageous.
Technical definitions create another obstacle. Autonomy may refer to navigation, target recognition, target selection, engagement or post-launch behavior. A system may be autonomous in one function and human-controlled in another.
Rapid technological change also complicates negotiations. Future systems may combine autonomous vehicles, machine learning, sensors and networked decision-making in ways that are difficult to classify today.
Verification and enforcement may be harder than agreement on general principles. Classified algorithms, proprietary software, encrypted communications, adaptive systems and distributed control networks can limit investigations and inspections.
Potential Effects of the UN Process
While negotiations continue, states may review procurement, testing and deployment policies. Military organizations could develop clearer rules for human authorization, target selection, operating boundaries, system failure, cybersecurity, command responsibility and civilian protection.
Future rules could also influence the design and marketing of autonomous systems. Potential requirements may involve auditability, human override functions, system documentation, target-recognition limits, reliability testing and compliance reviews.
The talks may clarify how existing international humanitarian law applies to autonomous weapons. Their eventual effect will depend on whether states adopt a binding treaty, a narrower framework, political commitments or no final agreement.
What Happens Next?
States must define their preferred regulatory approaches and may consult military officials, legal experts, technology specialists, humanitarian organizations and civil society groups.
Important developments may include:
- Formal UN meetings
- Negotiating sessions
- Working definitions
- Proposals for bans or restrictions
- Statements on human control
- Draft treaty language
- Verification proposals
- Accountability provisions
- National reporting commitments
- Progress toward the 2028 target
The most important indicator will be whether states move from general concern to precise obligations. Progress could include agreement on common terminology, shared principles, limits on specific uses, human oversight commitments, national weapons-review procedures and investigation requirements.
Conclusion: A Diplomatic Opening With Major Questions Ahead
States have agreed to pursue UN talks aimed at regulating autonomous weapons, or “killer robots,” with negotiations targeted for completion by 2028. The development places autonomous weapons higher on the international security and humanitarian agenda. Source 1
Major questions remain. Negotiators must define autonomy, determine what constitutes meaningful human control, assign accountability and decide which systems or uses should be prohibited. They must also design verification and enforcement mechanisms that work despite classified software and rapidly changing technology.
The agreement to negotiate is not a binding international rule. It does not create a global ban or guarantee a completed treaty by 2028. Its importance lies in opening a formal diplomatic process. The outcome will depend on whether states can reconcile humanitarian concerns with military interests and technological development.
Frequently Asked Questions
What are “killer robots”?
“Killer robots” is an informal term for autonomous weapons systems that may identify, select or attack targets with limited or no direct human control at the moment of engagement.
What did states agree to do?
States agreed to pursue United Nations talks aimed at regulating autonomous weapons systems. The negotiations are targeted for completion by 2028.
Does the agreement ban autonomous weapons?
No. The agreement advances negotiations. It does not create a global ban, require states to dismantle weapons or establish final binding rules.
What could the UN talks regulate?
The talks may address definitions, human control, prohibited uses, accountability, testing, deployment, transparency, verification and compliance.
Why are autonomous weapons controversial?
Critics argue that machines could make lethal decisions, harm civilians, weaken accountability and increase the risk of rapid escalation. Supporters may argue that some automated systems can provide defensive or military benefits.
Will there be a final agreement by 2028?
States have set a target of completing negotiations by 2028, but the target does not guarantee that governments will reach a final agreement on time.