One-sentence brief
Calling an entire drone autonomous hides where human judgment, model inference, communications, mission constraints, and force authority actually reside.
AI KILL CHAINS, KILL WEBS, AND AUTONOMOUS DECISION SYSTEMS
A function-specific analysis of AI, navigation, tracking, target recognition, task allocation, mesh networking, terminal guidance, and bounded engagement.
ANALYTICAL & SAFETY BOUNDARIES
Analytical boundary: A loitering munition combines sensing, persistence, and an effect in one platform, but that fact does not establish autonomous target selection, autonomous use of force, combat effectiveness, or one universal operating doctrine. Capability is not deployment; deployment is not autonomous engagement; navigation autonomy is not target-selection autonomy; manufacturer description is not independent operational verification; market forecasts are source claims rather than independently certified facts; a swarm demonstration is not proof of fielded unrestricted lethal autonomy. Public pages remain neutral, non-operational, synthetic where illustrative, and focused on evidence, governance, resilience, and defense.
Simulation safety boundary: Neutral, non-operational education. No real-person scoring, real targeting, weapons configuration, surveillance deployment, evasion, jamming instruction, or system-defeat guidance.
Source basis: Two exact owner-supplied WIP.76 reports.
ORIENTATION
Calling an entire drone autonomous hides where human judgment, model inference, communications, mission constraints, and force authority actually reside.
WORKING BRIEF
A platform may navigate, track, recognize, relay, or allocate tasks autonomously while a human retains target selection or engagement authority. Another system may engage a bounded target class after activation. These arrangements must be described separately.
Distributed drones can provide observations, relay data, maintain tracks, or advertise effect availability. A kill web composes possible mission pathways but does not erase the sequential obligations of identification, authorization, engagement, and assessment.
Contested links encourage more onboard processing, visual navigation, stored mission logic, and local coordination. This can preserve function while making supervision and correction harder.
COMPLETE DOSSIER
No related page is required to understand this record.
Terms are defined for this site’s evidence method, not as universal legal or clinical definitions.
| Function | What it may do | What it does not establish |
|---|---|---|
| Autonomous navigation | Adjust route or stabilize flight without continuous steering | Authority to select or attack a target |
| Visual or RF tracking | Maintain custody of an object or emitter | Positive identification or lawful target status |
| Automatic target recognition | Classify sensed features against a model or signature set | Legal distinction, proportionality, or unrestricted target selection |
| Task allocation | Assign search, relay, or observation roles among platforms | Permission to apply force |
| Mesh networking | Route data through alternative nodes | Trustworthiness, provenance, or command authority |
| Terminal guidance | Steer toward an operator-selected or predesignated object | Independent target nomination |
| Autonomous engagement within a bounded profile | Select and engage within preconfigured constraints | Open-ended authority or predictable performance in every environment |
| Human supervision | Monitor system activity and potentially intervene | Meaningful control when time, evidence, workload, or communications make intervention impossible |
RESEARCH EDITION
This page follows the public method for provenance, confidence, source independence, alternative accounts, limitations, review state, and visible correction.
CONTINUE