Autonomous Weapons Escalation

1. Scientific definition

Autonomous Weapons Escalation describes the systemic risk created when weapon systems are able to select, track, or engage targets with reduced human control, especially when such systems are integrated into military AI, surveillance, cyber, and command systems. The threat includes lethal autonomous weapon systems, machine-speed targeting, algorithmic identification, human-machine decision chains, automated defense systems, drone swarms, digital targeting infrastructures, and escalation pathways that may develop faster than political leaders can understand or restrain (ICRC, 2023; SIPRI, 2017).

The term should not be confused with every remotely operated drone or every digital tool used by militaries. A remotely piloted aircraft still depends on direct human command for target engagement. Autonomous weapons raise a different problem: human judgment may be displaced, compressed, or transformed by systems that process sensor data and act at speeds and scales beyond ordinary deliberation. The ICRC describes autonomous weapons as an immediate cause of humanitarian concern and recommends new, international legally binding rules (ICRC, 2023); in October 2023 the UN Secretary-General and the ICRC President jointly called for such negotiations to be concluded by 2026 (UN Secretary-General and ICRC, 2023).

Autonomous Weapons Escalation is also distinct from artificial intelligence in general. The systemic concern is not software alone, but the incorporation of machine decision-making into the use of force. SIPRI's mapping of autonomy in weapon systems and successive UN discussions on lethal autonomous weapon systems document the expansion of autonomy in existing weapon systems, which in 2017 was concentrated mainly in defensive systems, and continuing international concern about its further development (SIPRI, 2017; UN CCW GGE, 2014 to 2026).

A model threshold for this threat cannot be treated as an empirical tipping point, a physical boundary, or a deterministic forecast. No direct observed threshold links autonomous weapons deployment to systemic destabilization. The Apocalypse Clock treats the threshold as a normalized systemic-risk anchor, reflecting warnings that autonomous weapons could destabilize deterrence, enable machine-speed escalation, create attribution problems, and increase civilian harm at scale (ICRC, 2023; UN CCW GGE, 2014 to 2026).

2. Empirical evidence base

The scale evidence is substantial because autonomous weapons could affect many conflicts, not only one battlefield. The ICRC describes autonomous weapons as an immediate humanitarian concern, with risks for civilians, combatants, and compliance with international humanitarian law. The scale assessment is therefore high, though not maximal, because the evidence supports large and widespread harm rather than a direct claim of civilization-ending damage from every deployment (ICRC, 2023).

Urgency is anchored in the ICRC's call for new, international legally binding rules and in the joint UN Secretary-General and ICRC appeal to conclude negotiations on them by 2026. This indicates that the problem is already present in policy and legal debate, rather than being confined to future speculation. The urgency also comes from the fact that military systems can be developed, tested, exported, and normalized before international law catches up (ICRC, 2023; UN Secretary-General and ICRC, 2023).

Acceleration evidence comes from the expanding development of autonomy in weapon systems. SIPRI's 2017 mapping study documents a steady expansion of autonomy in weapon systems, then concentrated mainly in defensive systems, and successive UN CCW discussions show continuing concern about its further development. This supports a fast acceleration reading, although these sources predate much recent battlefield use and the evidence does not provide a single global registry of all autonomous weapons in use (SIPRI, 2017; UN CCW GGE, 2014 to 2026).

The growth-pressure evidence is proxy-based. Industry forecasts for autonomous and automated military weapons markets cluster around compound annual growth of roughly 7 to 13 percent. In the Apocalypse Clock model, this market and capability growth is used only as a diffusion-pressure proxy. It is not a direct measurement of systemic security risk-growth or annual escalation probability (Allied Market Research, 2022; TechSci Research, 2025).

Interdependence is one of the most important dimensions of the threat. SIPRI's work on military AI emphasizes that autonomy is being integrated into wider military and security technologies, including conventional, cyber and nuclear force-related systems. Failures or escalation in autonomous weapons systems can therefore interact with conventional conflict, cyber operations, and nuclear decision environments (SIPRI AI Programme).

Irreversibility is moderate and conditional. The UN Secretary-General and the ICRC warn that autonomous weapons may lower the threshold for engaging in conflict, and state submissions summarized in the UN Secretary-General's 2024 report warn of an arms race and of proliferation. Yet the evidence also preserves an important boundary: binding rules, prohibitions, and regulation could still limit deployment and prevent some pathways from becoming locked in (UN Secretary-General and ICRC, 2023; UN Secretary-General, 2024; UN CCW GGE, 2014 to 2026).

Governance failure is already visible in the treaty process. After more than a decade of CCW discussions on lethal autonomous weapon systems (informal meetings from 2014, a Group of Governmental Experts since 2017, with its current mandate running to 2026), there is still no legally binding global treaty. UN General Assembly debates and civil-society campaigns show growing support for regulation, but the legal regime remains incomplete (UN CCW GGE, 2014 to 2026; Stop Killer Robots).

3. Mechanism of systemic destabilization

Autonomous Weapons Escalation destabilizes security systems by compressing the time between detection, interpretation, and force. Human commanders traditionally operate through procedures that include judgment, verification, hesitation, and responsibility. Autonomous systems can shorten that chain. ICRC warnings about human control and humanitarian risk support concern that speed and distance from direct human judgment may change the moral and strategic structure of force (ICRC, 2023).

The first mechanism is machine-speed escalation. If autonomous systems identify threats and respond rapidly, opposing systems may be designed to respond even faster. That dynamic can create pressure to delegate more decisions to machines. The threat is not that every autonomous weapon makes war automatic; the risk is that crisis stability weakens when military organizations believe slower human deliberation has become a disadvantage (ICRC, 2023; SIPRI, 2017).

The second mechanism is threshold lowering. Weapons that reduce risks to the deploying force can make the use of force appear less politically costly. The UN Secretary-General and the ICRC warn that autonomous weapons may lower the threshold for engaging in conflict, and several states made the same point in submissions to the UN Secretary-General. This pathway is systemic because many states and non-state actors may adjust doctrine, procurement, and crisis behavior once autonomous systems become normalized (UN Secretary-General and ICRC, 2023; UN Secretary-General, 2024).

The third mechanism is attribution and accountability failure. When targeting decisions are distributed across sensors, software, operators, commanders, vendors, training data, and battlefield conditions, responsibility becomes harder to assign. ICRC concerns about legally binding rules reflect this problem: international humanitarian law depends on human judgment, responsibility, and review, not only on technical performance (ICRC, 2023).

The fourth mechanism is integration with surveillance and command systems. SIPRI’s military AI programme highlights the integration of artificial intelligence into wider military and security systems. Autonomous weapons do not operate in isolation. They may draw on surveillance feeds, intelligence databases, cyber infrastructure, communications networks, and command platforms. A failure in one layer can alter targeting or escalation behavior in another (SIPRI AI Programme).

The fifth mechanism is cross-domain escalation. Autonomous weapons can interact with cyber operations, conventional military crises, and nuclear signaling because command systems and warning systems are increasingly digitized. The allowed evidence supports this as an interdependence pathway, not as a claim that autonomous weapons automatically cause nuclear escalation. The concern is narrower: military AI and autonomy can make crises faster, more opaque, and more difficult to interpret (SIPRI AI Programme; UN CCW GGE, 2014 to 2026).

The sixth mechanism is arms-race entrenchment. Once major powers and regional rivals invest in autonomous capabilities, others may feel pressure to follow. State submissions summarized in the UN Secretary-General's 2024 report warn that lethal autonomous weapons systems might become the objects of an arms race. Market forecasts indicating rapid growth in autonomous and automated military weapons provide a proxy for this diffusion pressure, though they do not measure actual systemic security harm directly (UN Secretary-General, 2024; Allied Market Research, 2022; TechSci Research, 2025).

The seventh mechanism is civilian harm at scale. Autonomous weapons may be deployed across many conflicts and environments, including urban areas, borders, maritime zones, and air-defense contexts. The ICRC frames the issue as an immediate humanitarian concern because target selection and engagement by autonomous systems may produce unacceptable risks for civilians and combatants. The scale of harm depends on design, doctrine, human control, environment, and legal constraints (ICRC, 2023).

The eighth mechanism is governance lag. CCW discussions have continued for years without producing a legally binding global treaty. That lag matters because military procurement and battlefield experimentation can advance faster than legal consensus. Stop Killer Robots campaign updates and UN discussions show growing political and civil-society concern, but the absence of binding rules leaves a gap between recognized risk and enforceable constraint (UN CCW GGE, 2014 to 2026; Stop Killer Robots).

The final mechanism is systemic security uncertainty. Autonomous weapons can make conflict harder to interpret: Was an attack deliberate, automated, accidental, spoofed, or triggered by faulty data? In a tense crisis, uncertainty itself can become destabilizing. The Apocalypse Clock treats its numerical scores as normalized inputs, not direct empirical measurements. The evidence supports a high systemic reading: autonomy in weapon systems is expanding, legal controls remain incomplete, and escalation pathways are plausible despite the absence of an empirical destabilization threshold (ICRC, 2023; SIPRI, 2017; SIPRI AI Programme; UN CCW GGE, 2014 to 2026).

4. Sources used

ICRC, 2023: Autonomous weapons (topic page, first published 2023, updated 2025), ICRC website.
URL: https://www.icrc.org/en/law-and-policy/autonomous-weapons
Supports: Evidence that autonomous weapons are an immediate cause of humanitarian concern, that the ICRC recommends new, international legally binding rules, and that human control and civilian protection are central risks.

SIPRI, 2017: Mapping the Development of Autonomy in Weapon Systems, SIPRI report by Vincent Boulanin and Maaike Verbruggen.
URL: https://www.sipri.org/publications/2017/policy-reports/mapping-development-autonomy-weapon-systems
Supports: Evidence on the state of autonomy in existing weapon systems as of 2017, then concentrated mainly in defensive systems, and on the drivers of further autonomy.

SIPRI, n.d.: Artificial intelligence (research area, Armament and Disarmament programme), SIPRI website.
URL: https://www.sipri.org/research/armament-and-disarmament/emerging-military-and-security-technologies/artificial-intelligence
Supports: Evidence that AI may find uses in conventional, cyber and nuclear force-related systems, creating interdependence with other security domains.

UN CCW Group of Governmental Experts on lethal autonomous weapon systems, 2014 to 2026: process tracker, Digital Watch Observatory (DiploFoundation).
URL: https://dig.watch/processes/gge-laws
Supports: Evidence on repeated international discussions of lethal autonomous weapon systems, governance gaps, unresolved legal questions, and the absence of a binding global treaty.

Allied Market Research, 2022: Autonomous Weapons Market Expected to Reach $30.16 Billion by 2030 (press release; 10.4 percent CAGR, 2021 to 2030), alliedmarketresearch.com.
URL: https://www.alliedmarketresearch.com/press-release/autonomous-weapons-market.html
Supports: Proxy-based evidence for autonomous-weapons market and capability growth, used cautiously as a diffusion-pressure indicator rather than a direct measure of systemic risk.

TechSci Research, 2025: Autonomous Military Weapons Market to Grow 9.72% CAGR through 2029, techsciresearch.com.
URL: https://www.techsciresearch.com/news/17897-autonomous-military-weapons-market.html
Supports: A second market forecast (9.72 percent CAGR through 2029) inside the cited growth range, used only as a diffusion-pressure proxy.

UN Secretary-General and ICRC, 2023: Joint call by the United Nations Secretary-General and the President of the International Committee of the Red Cross for States to establish new prohibitions and restrictions on Autonomous Weapon Systems, 5 October 2023, ICRC.
URL: https://www.icrc.org/en/document/joint-call-un-and-icrc-establish-prohibitions-and-restrictions-autonomous-weapons-systems
Supports: The call to conclude negotiations on a legally binding instrument by 2026, and the warning that autonomous weapons may lower the threshold for engaging in conflict and contribute to global instability.

UN Secretary-General, 2024: Lethal autonomous weapons systems, Report of the Secretary-General, A/79/88, UN General Assembly.
URL: https://docs-library.unoda.org/General_Assembly_First_Committee_-Seventy-Ninth_session_(2024)/A-79-88-LAWS.pdf
Supports: State views that lethal autonomous weapons systems might become the objects of an arms race, could lower the threshold for the use of force, and could proliferate to non-state actors.

Stop Killer Robots, campaign updates on autonomous weapons.
URL: https://www.stopkillerrobots.org/
Supports: Evidence on civil-society calls for prohibitions and regulation, plus the continuing gap between political concern and binding legal control.