Intent-Bound Defense Engagement

What This Application Specifies

Operator intent enters the architecture as a credentialed declaration with explicit, machine-readable fields: intended objective (the military task the engagement serves), intended scope (geographic boundary, temporal window, target-class taxonomy), intended rules of engagement (RoE profile, weapons-release authority, escalation ladder), and intended termination conditions (abort triggers, mission-complete criteria, safe-state fallbacks). The intent record carries the credential of the issuing authority and admits through composite admissibility before any engagement actuator is permitted to act.

Intent-authority composition structures map to the multi-echelon reality of defense command. Tactical commanders issue tactical intent bounded by the theater commander's intent; theater commanders operate within national-command-authority intent; national-command-authority intent operates within the legal framework of the law of armed conflict and the deploying state's Article 36 weapons-review determinations. The architecture represents these layers as nested, credentialed intent envelopes, so that a lower-echelon engagement can never structurally exceed the bounds set by a higher-echelon authority, and so that compliance with DoD 3000.09's requirement for "appropriate levels of human judgment over the use of force" becomes a property of the system rather than an aspiration laid over it.

The specification also distinguishes intent from command. A command is an instruction to the system; intent is the prior, persistent declaration that gives the command meaning and bounds its lawful execution. The architecture stores both, links them, and refuses to execute a command that admits no intent envelope or that exceeds the envelope it claims. The distinction matters because the law of armed conflict places its substantive obligations on the human judgment behind the act, not on the act itself; an architecture that records only commands cannot speak to that judgment, while an architecture that records intent envelopes makes the judgment auditable as a first-class object.

Why It Matters Operationally

Current defense-autonomy systems face a structural intent gap. Operators issue commands that the system executes; the relationship between operator intent and system behavior is implementation-dependent, encoded in mission-planning software, target-decision aids, and human-machine interfaces that vary across program offices and platform vendors. When investigators or weapons-review boards ask after the fact what the operator intended, the answer is reconstructed from logs, interviews, and inference rather than retrieved from a structurally-recorded artifact.

This gap matters legally as well as operationally. AP I Article 48 (the basic rule of distinction), Article 51 (protection of the civilian population), and Article 57 (precautions in attack) all demand that those who plan and decide on attacks act on a particular state of mind: belief that the target is a military objective, judgment that expected civilian harm is not excessive in relation to anticipated military advantage, and choice of means and methods to minimize incidental harm. The ICRC's framework on autonomy in weapon systems, and the broader CCW Group of Governmental Experts work, repeatedly return to the same point: the human judgment that the law requires must be exercised meaningfully and traceably, not delegated by default to whatever the autonomy stack happened to do.

Intent-bound execution closes the gap by producing structural support for that judgment. The intent is declared at the appropriate echelon; admissibility evaluates each candidate engagement against the intent envelope; execution proceeds only within scope; deviations from intent fail admissibility structurally rather than relying on the autonomy stack to refuse on its own initiative. The same record that authorizes the engagement is the record that, after the fact, allows a weapons-review board, a command investigation, or an external accountability body to reconstruct what was intended and to compare it against what occurred.

How It Composes With the Domain

Each engagement actuation admits against the active intent envelope. A target nomination arrives with cross-modality observations (radar track, electro-optical confirmation, signals correlation); the admissibility evaluation checks not only the technical confidence in identification but also whether the candidate target falls within the declared target-class taxonomy, whether the geographic location lies within the declared geographic scope, and whether the engagement window falls within the declared temporal scope. A target that satisfies the technical thresholds but lies outside the intent envelope does not proceed, and the refusal is itself a credentialed event admissible into after-action review.

Stage-gated commitment composes naturally with intent. Lock, track, weapons-free, and weapons-release each draw their authority from the intent envelope, and each stage emits a credentialed transition. The discipline that NATO targeting doctrine and U.S. joint targeting cycles (find, fix, track, target, engage, assess) describe procedurally becomes a sequence of admissibility checks against a structured intent record, with the same record carrying through to post-strike battle-damage assessment.

Adversarial intent manipulation surfaces as credentialed integrity events. Forced-intent attempts (an adversary attempting to cause an authority to issue a permissive intent under coercion or deception), intent-spoofing (forged intent records or replayed envelopes), and intent-extraction (attempts to enumerate the system's intent envelope to plan around it) each enter the architecture as named, credentialed event types. The architecture does not claim to prevent every such attempt; it claims that attempts leave structural traces that downstream review can act on.

Composition with allied and coalition operations follows the same pattern. Coalition rules of engagement, national caveats, and shared targeting boards all map to multi-authority intent envelopes whose intersection defines the lawful engagement space, with each contributing authority retaining a credentialed view of what it authorized and what it did not.

The five-property chain disclosed in U.S. Provisional Application No. 64/049,409 governs each engagement end-to-end: the issuing commander's intent record is an authority-credentialed observation, the cross-modality target nominations are evidentially weighted against the active envelope, the admissibility evaluation produces a composite admissibility decision over the joined evidence, the resulting fire-control or refusal action is a governed actuation bounded by that decision, and the entire transition is recorded with lineage-recorded provenance whose chain closes recursively when post-strike assessment re-enters the same envelope as a fresh credentialed observation. No engagement step exits the chain; no audit reconstructs from outside it.

What This Enables

Defense operations gain structurally-supported meaningful-human-control. The phrase, which has carried the diplomatic weight of the autonomy-in-weapons debate for more than a decade, acquires a concrete referent: the intent envelope that bounds execution, the admissibility evaluation that enforces it, and the credentialed record that preserves it. Operating authorities gain audit-grade intent reconstruction, sufficient for command investigations, inspector-general inquiries, and the after-action obligations that AP I Article 87 places on commanders.

Article 36 weapons-review processes gain a more tractable object of review. A reviewing authority can examine the intent envelopes a system is designed to admit, the admissibility logic that enforces them, and the integrity events the system surfaces, rather than attempting to certify an opaque autonomy stack against an open-ended set of imagined uses. ICRC engagement, CCW deliberation, and domestic legislative oversight gain the same tractability: the question shifts from "is the system lawful in the abstract" to "are the intent envelopes the system admits, and the enforcement around them, adequate to the obligations the deploying state has accepted."

The architecture also supports doctrine evolution. As meaningful-human-control doctrine refines, as intent-formulation tools mature, as intent-extraction protections improve, and as new categories of autonomous capability emerge, the architecture admits the changes through declared specification rather than through reimplementation of the underlying system. The discipline is durable; the doctrine on top of it can move.

Adoption Path and Interface to Existing Doctrine

Adoption of the intent-bound posture does not require rewriting joint doctrine, replacing fielded autonomy stacks, or re-litigating the unresolved policy questions around lethal autonomous weapon systems. It requires interposing the intent envelope as a structural object between the authority that issues it and the actuator that consumes it, and then progressively migrating admissibility checks from implicit, per-vendor implementations to evaluations against the envelope. Existing artifacts, mission-type orders, fragmentary orders, air tasking orders, special-instructions documents, target-engagement authorities, already encode intent in human-readable form; the migration path captures their structure as credentialed records rather than discarding their content.

The interface to Article 36 reviews is similarly incremental. A reviewing authority can begin by certifying the intent envelope schema and the admissibility logic that enforces it, treating the underlying autonomy stack as a separately-governed component whose outputs admit only within the envelope. As confidence in the envelope discipline grows, more of the engagement authorization can be expressed as envelope structure rather than as out-of-band policy, and the review burden migrates accordingly. The same pattern serves coalition rules of engagement, where each contributing nation's caveats become declared envelope constraints whose intersection the architecture computes and enforces, rather than a coordination problem that staffs have to resolve verbally before each operation.

Closing Note

The intent-bound posture does not resolve the substantive moral and legal questions of autonomous engagement. It does something more modest and more useful: it gives those questions a place to land in the architecture, so that the answers a society reaches, through doctrine, treaty, and law, can be enforced as structural properties of the systems that act in its name rather than as policies layered over systems that were never built to carry them.