Skip to main content
← All campaigns

LAWS Regulation: Autonomous Weapons Governance in Structural Collapse

stalled
22
17 entriessince 2014-01-01

Tracks the governance landscape for lethal autonomous weapons systems (LAWS): UN CCW GGE negotiations, REAIM voluntary frameworks, national legislation (US AI Guardrails Act), EU AI Act military exemption, China's "cumulative criteria" definitional strategy, and the accountability void exposed by Iran war AI deployments. Key dynamics: governance inversely correlated with capability — REAIM endorsements declining, binding treaty stalled, major powers legislating to enable. 7th CCW Review Conference and Secretary-General's end-2026 deadline make next 8 months decisive.

lawsgovernanceautonomous_weaponsmilitary_aiarms_control

Signals

Timeline

2026-04-13
capability
Anthropic withholds Claude Mythos Preview; confines it to 40-org Project Glasswing after autonomous zero-day discovery

Anthropic declined to release Claude Mythos Preview after internal and third-party testing showed autonomous discovery and exploit generation for zero-day vulnerabilities in open-source codebases at…

ai_safetycybersecuritydual_useanthropicgovernance
2026-04-10
assessment
ASSESSMENT: LAWS regulation in structural collapse — governance inversely correlated with capability

ASSESSMENT: The LAWS governance landscape in April 2026 reveals an inverse correlation: as autonomous weapons capabilities accelerate, governance frameworks are weakening. Evidence across all tracks:…

assessmentstructural_collapseinverse_correlationgovernance_gapdecisive_period
2026-04-01
policy
Slotkin AI Guardrails Act (S.4113): restricts autonomous lethal force — but waiver allows override

Senator Elissa Slotkin (D-MI) introduces AI Guardrails Act of 2026 (S.4113, March 17). Primary purpose is RESTRICTING Pentagon AI use: prohibits autonomous lethal force without human authorization,…

slotkinai_guardrails_actautonomous_weaponslegislationwaiveranthropic
2026-03-27
analysis
China's "human oversight with Chinese characteristics": cumulative criteria that exclude real-world systems

Lieber Institute analysis (Mako, March 27): China's position on LAWS regulation employs a distinctive "cumulative criteria" framework that appears restrictive but is functionally permissive.…

chinaccwcumulative_criteriahuman_oversightdefinitional_strategyliterature
2026-03-19
deployment
Anduril starts Fury combat drone production — Arsenal-1 factory operational, Replicator scaling

Anduril Industries begins production of YFQ-44 Fury "loyal wingman" high-speed combat drones at its $1B Arsenal-1 autonomous systems manufacturing campus in Columbus, Ohio. Factory will employ 4,000+…

andurilarsenal_1furyreplicatordawgindustrial_base
2026-03-15
diplomacy
CCW GGE LAWS 2026 first session in Geneva — 156 states back UNGA resolution, US/Russia vote against

The CCW Group of Governmental Experts on LAWS held its first 2026 session in Geneva (March). Second session planned August-September 2026 with goal of finalizing a consensus report for the 7th CCW…

ccwggegenevaunga_resolutionbinding_treatytwo_tier
2026-03-14
deployment
Pentagon awards Anduril $20B contract for AI battlefield network ('Arsenal-1'...

Pentagon awards Anduril $20B contract for AI battlefield network ('Arsenal-1' megafactory in Ohio, operational by mid-2026). Combined with Pentagon's 'AI-first' Acceleration Strategy and the…

pentagonandurilprocurementautonomous_systemsreplicator
2026-03-10
assessment
Iran war exposes accountability void: AI strike on Minab school (165 dead), no framework to assign blame

The Iran war has become the first large-scale test case for AI weapon accountability — and the frameworks are failing. Bulletin of Atomic Scientists (Goudarzi, March 10): uncertainty over whether…

accountabilityminabcivilian_harmiran_warlegal_gapmeaningful_human_control
2026-03-01
paper
Matosek & Prokopowicz: generative AI in combat systems outpaces legal norms

Matosek & Prokopowicz: generative AI in combat systems outpaces legal norms

papergenerative_ai
2026-02-05
diplomacy
REAIM 2026 Summit: US, China, Russia, Israel refuse to sign — endorsements drop from 60+ to 39

Third REAIM Summit (A Coruña, Spain, Feb 4-5) produces "Pathways for Action" document with 20 specific measures across three areas: AI impact on international peace/security, responsible AI…

reaimsummitspainvoluntary_governancedeclining_support
2026-02-01
policy
EU AI Act explicitly exempts military — "meaningful human control" unenforceable beyond funding conditions

The EU AI Act, the world's most comprehensive AI regulation, contains an express exemption removing military AI from its risk-based framework. The European Defence Fund mandates "meaningful human…

euai_actmilitary_exemptioneuropean_defence_fundmeaningful_human_control
2026-01-31
assessment
Peer-reviewed ACF analysis identifies two competing coalitions in military AI...

Peer-reviewed ACF analysis identifies two competing coalitions in military AI governance: 'tight regulation' (UN GGE/LAWS, binding treaty) vs 'responsible use' (REAIM Summits, US-led Political…

assessmentgovernancecoalitionsreaim
2026-01-01
paper
Jankovic et al.

Jankovic et al.: comprehensive risk analysis — algorithmic bias, unintended escalation, accountability gaps

paperrisk_analysis
2025-01-01
paper
Bayar: IHL has no specific provisions for AWS, legal frameworks need adaptation

Bayar: IHL has no specific provisions for AWS, legal frameworks need adaptation

paperihl
2024-01-01
paper
Simmons-Edler et al.

Simmons-Edler et al. (arXiv): AWS reduce political cost of aggression, increase conflict risk

paperconflict_risk
2018-01-01
assessment
Global pact on autonomous lethal weapons predicted to fail (Oxford Analytica)

Global pact on autonomous lethal weapons predicted to fail (Oxford Analytica)

assessmentfailure
2014-01-01
milestone
UN CCW (Convention on Certain Conventional Weapons) discussions on LAWS begin

UN CCW (Convention on Certain Conventional Weapons) discussions on LAWS begin

unccw