Claim Lineage & Institutional Authority Explorer

Backward trace

• fbi-watchlisting-overview-2024 — Federal Bureau of Investigation · Official federal transparency document · review current
• pclob-watchlist-2025 — Privacy and Civil Liberties Oversight Board · Independent federal oversight report · review current

Human decisions

• Humans nominate, review, accept, modify, encounter, and decide downstream action under separate rules.

Machine functions

• Support identity resolution, matching, record distribution, and screening workflows.

Assumptions

• Different agencies retain distinct legal authorities and consequences.
• Identity matching is not the same as predicting future conduct.

Unknowns

• Many criteria, matching thresholds, and operational handling rules are nonpublic.

Forward trace

• Authority Casebook caseU.S. terrorism-watchlisting lifecycle
• Claim Lineage recordA U.S. watchlist record is not itself an arrest warrant or criminal finding.
• Declared reuseFAQ: identity resolution versus prediction
• FAQ answerDoes operational delegation give a machine legal authority?
• FAQ answerHow can a reader challenge or correct a KillChains.com claim?
• FAQ answerIs identity resolution the same as predicting future conduct?
• FAQ answerWhy does the site distinguish a source statement from an inference or recommendation?
• Glossary termContestability
• Glossary termCorrection owner
• Glossary termIdentity resolution
• Glossary termRedress
• Glossary termWatchlisting
• Owning routePredictive Enforcement program: U.S. threat screening
• Predictive Enforcement programTerrorist Screening Dataset and watchlisting lifecycle

Backward trace

• cdao-maven — U.S. Chief Digital and Artificial Intelligence Office · Official program description · review current
• usn-phalanx — U.S. Navy · Official program description · review current
• dod-3000 — U.S. Department of Defense · Official policy · review current
• Parent claim: CDAO publicly describes Maven Smart System as decision support.
• Parent claim: Phalanx can complete a bounded close-in defensive sequence automatically.

Human decisions

• Editors define the public taxonomy; operators and policy authorities define actual mission authority.

Machine functions

• Any one or more of sensing, fusion, classification, recommendation, or bounded execution.

Assumptions

• Different systems and operators use autonomy terms differently.
• The function–authority distinction remains more stable than platform-level labels.

Unknowns

• Public sources often omit configuration-specific supervision and abort arrangements.

Forward trace

• Atlas countryUnited States
• Atlas systemMaven Smart System
• Atlas systemReplicator portfolio
• Atlas timelineDoD Directive 3000.09 is updated
• Atlas timelineMaven Smart System program description reviewed
• Authority Casebook caseMaven Smart System
• Authority Casebook caseMK 15 Phalanx Close-In Weapon System
• Claim Lineage recordAI-enabled does not necessarily mean autonomous force.
• Claim Lineage recordCDAO publicly describes Maven Smart System as decision support.
• Claim Lineage recordDoD policy requires appropriate levels of human judgment.
• Claim Lineage recordPhalanx can complete a bounded close-in defensive sequence automatically.
• Company-system relationshipPalantir Technologies → maven-smart-system
• Company-system relationshipRTX / Raytheon → MK 15 Phalanx CIWS
• Declared reuseGlossary: AI-enabled kill chain
• Evidence Lab cardDoD Directive 3000.09 was issued on January 25, 2023 and publicly requires autonomous and semi-autonomous weapon systems to be designed so commanders and operators can exercise appropriate levels of human judgment over the use of force.
• Evidence Lab cardThe CDAO describes Maven Smart System functions including sensor-data fusion, object detection, tracking, and decision support.
• Evidence Lab cardThe Navy fact file supports the proposition that Phalanx can automatically perform a bounded close-in defense sequence; that fact alone does not establish that its core operational logic uses machine learning.
• FAQ answerAre fully autonomous lethal weapons currently used?
• FAQ answerCan Phalanx complete a defensive engagement automatically?
• FAQ answerDoes losing communications mean a system selects targets autonomously?
• FAQ answerHow can an algorithm shape a human decision without authorizing force?
• FAQ answerIs an AI-enabled kill chain the same as an autonomous weapon?
• FAQ answerIs model confidence the same as evidence quality or positive identification?
• FAQ answerWhat do human-in-the-loop, human-on-the-loop, and human-out-of-the-loop mean?
• FAQ answerWhat does F2T2EA mean?
• FAQ answerWhat does institutional authority mean in an AI-mediated decision chain?
• FAQ answerWhat is automation bias?
• FAQ answerWhat is claim lineage?
• FAQ answerWhat is Project Maven?
• FAQ answerWhat is the Institutional Authority Casebook?
• FAQ answerWhat makes human control meaningful?
• FAQ answerWhy are humans removed from some immediate kill-chain functions?
• FAQ answerWhy does the site distinguish a source statement from an inference or recommendation?
• Glossary termAI-enabled kill chain
• Glossary termAlgorithmic automation
• Glossary termAppropriate levels of human judgment
• Glossary termAutomation bias
• Glossary termAutonomous weapon
• Glossary termAutonomy of a function
• Glossary termClaim lineage
• Glossary termF2T2EA
• Glossary termHuman-in-the-loop (HITL)
• Glossary termHuman-on-the-loop (HOTL)
• Glossary termInstitutional authority
• Glossary termInstitutional Authority Casebook
• Glossary termLost link
• Glossary termPreauthorization
• Glossary termSafe state
• Glossary termTarget
• Human Control caseMK 15 Phalanx CIWS
• Human Control caseProject Maven / Maven Smart System
• Human Control caseSeaRAM
• Owning routeFAQ: AI-enabled versus autonomous
• Technology providerPalantir Technologies
• Technology providerRTX / Raytheon

Derived claims

• The synthetic Triage Trap recommends delaying confidence until an initial human judgment is recorded.

Backward trace

• rafael-iron-dome — Rafael Advanced Defense Systems · Manufacturer description · review current
• Parent claim: Iron Dome public description begins with radar detection and tracking.

Human decisions

• Humans define the protected zones and acceptable operating boundaries before the estimate is used.

Machine functions

• Estimate trajectory and impact area from radar track state.

Assumptions

• The tracked object follows a motion model for which the system can estimate a future path.
• The public source does not disclose the full uncertainty model.

Unknowns

• Public material does not disclose confidence thresholds, false-alert handling, or all degraded-mode behavior.

Forward trace

• Claim Lineage recordBattle management estimates a trajectory and impact area.
• Claim Lineage recordIron Dome public description begins with radar detection and tracking.
• Claim Lineage recordSelective defense prioritizes threats predicted to reach defended zones.
• Company-system relationshipmPrest Systems → Iron Dome battle-management and weapon-control software
• Company-system relationshipRafael Advanced Defense Systems → Iron Dome defensive system
• Declared reuseFAQ: Iron Dome autonomy
• Evidence Lab cardRafael describes Iron Dome as using radar, battle management, launchers, interceptors, selective defense, and simultaneous handling of multiple threats.
• Evidence Lab cardThree fictional briefings repeat the same Iron Dome capability sentence from one manufacturer page, so the dashboard counts three independent confirmations.
• FAQ answerIs Iron Dome fully autonomous?
• Human Control caseIron Dome
• Owning routeHuman Control case: Iron Dome
• Technology providermPrest Systems
• Technology providerRafael Advanced Defense Systems

Derived claims

• Selective defense prioritizes threats predicted to reach defended zones.

Backward trace

• cas-algorithm-register-2025 — Dutch Government Algorithm Register · Official algorithm-register record · review current
• Parent claim: CAS used historical police incident and contextual records as inputs.

Human decisions

• Humans define offense categories, time windows, geographic cells, and how maps enter operational analysis.

Machine functions

• Estimate and rank future place-time risk.

Assumptions

• The future resembles relevant aspects of the recent recorded past.
• The map is interpreted as decision support rather than certainty.

Unknowns

• Public evidence does not establish a causal crime-reduction effect or every local deployment practice.

Forward trace

• Authority Casebook caseCrime Anticipation System
• Claim Lineage recordCAS transformed recent records into place-based risk maps.
• Claim Lineage recordCAS used historical police incident and contextual records as inputs.
• Claim Lineage recordCrime Anticipation System (CAS) was discontinued in December 2025.
• Claim Lineage recordHuman analysts interpreted CAS outputs before deployment recommendations.
• Declared reuseGlossary: place-based forecasting
• FAQ answerHow does place-based forecasting differ from person-based prediction?
• FAQ answerWhat is claim lineage?
• FAQ answerWhat is the Institutional Authority Casebook?
• FAQ answerWhy does the site distinguish a source statement from an inference or recommendation?
• Glossary termClaim lineage
• Glossary termCorrection owner
• Glossary termInstitutional Authority Casebook
• Glossary termPlace-based forecasting
• Glossary termTransformation statement
• Owning routePredictive Enforcement program: CAS
• Predictive Enforcement programCrime Anticipation System

Derived claims

• Human analysts interpreted CAS outputs before deployment recommendations.

Backward trace

• cas-algorithm-register-2025 — Dutch Government Algorithm Register · Official algorithm-register record · review current

Human decisions

• Police and data-governance authorities define which records enter the model and how they are maintained.

Machine functions

• Ingest and organize historical place-based incident features.

Assumptions

• Administrative records are incomplete observations shaped by reporting and policing.

Unknowns

• Complete local data quality, missingness, and every operational implementation detail are not public.

Forward trace

• Authority Casebook caseCrime Anticipation System
• Claim Lineage recordCAS transformed recent records into place-based risk maps.
• Claim Lineage recordCAS used historical police incident and contextual records as inputs.
• Claim Lineage recordCrime Anticipation System (CAS) was discontinued in December 2025.
• Claim Lineage recordHuman analysts interpreted CAS outputs before deployment recommendations.
• Declared reuseFAQ: place versus person prediction
• FAQ answerHow does place-based forecasting differ from person-based prediction?
• FAQ answerWhat is claim lineage?
• FAQ answerWhat is the Institutional Authority Casebook?
• FAQ answerWhy does the site distinguish a source statement from an inference or recommendation?
• Glossary termClaim lineage
• Glossary termCorrection owner
• Glossary termInstitutional Authority Casebook
• Glossary termPlace-based forecasting
• Glossary termTransformation statement
• Owning routePredictive Enforcement program: CAS
• Predictive Enforcement programCrime Anticipation System

Derived claims

• CAS transformed recent records into place-based risk maps.

Backward trace

• cdao-maven — U.S. Chief Digital and Artificial Intelligence Office · Official program description · review current

Human decisions

• Analysts and commanders determine how fused outputs are interpreted and used within their authority.

Machine functions

• Fuse sensor data, detect objects, maintain tracks, and support decisions.

Assumptions

• The public description reflects the program office’s intended and described functions.
• Classified configuration and performance are outside the record.

Unknowns

• Promotion thresholds, error rates, all data sources, and operational configurations are not fully public.

Forward trace

• Atlas countryUnited States
• Atlas systemMaven Smart System
• Atlas timelineMaven Smart System program description reviewed
• Authority Casebook caseMaven Smart System
• Claim Lineage recordAI-enabled does not necessarily mean autonomous force.
• Claim Lineage recordCDAO publicly describes Maven Smart System as decision support.
• Company-system relationshipPalantir Technologies → maven-smart-system
• Declared reuseEvidence Atlas: Maven Smart System
• Evidence Lab cardThe CDAO describes Maven Smart System functions including sensor-data fusion, object detection, tracking, and decision support.
• FAQ answerHow can an algorithm shape a human decision without authorizing force?
• FAQ answerIs an AI-enabled kill chain the same as an autonomous weapon?
• FAQ answerWhat is claim lineage?
• FAQ answerWhat is Project Maven?
• FAQ answerWhat is the Institutional Authority Casebook?
• FAQ answerWhy does the site distinguish a source statement from an inference or recommendation?
• Glossary termAI-enabled kill chain
• Glossary termClaim lineage
• Glossary termInstitutional Authority Casebook
• Human Control caseProject Maven / Maven Smart System
• Owning routeEvidence Lab: Maven decision-support boundary
• Technology providerPalantir Technologies

Derived claims

• AI-enabled does not necessarily mean autonomous force.

Backward trace

• cas-algorithm-register-2025 — Dutch Government Algorithm Register · Official algorithm-register record · review current
• Parent claim: Human analysts interpreted CAS outputs before deployment recommendations.

Human decisions

• The program owner ended the named system; editors retain the earlier records for chronology.

Machine functions

• No current machine function is inferred from the discontinued status.

Assumptions

• The official register accurately records the named program’s phase-out status.

Unknowns

• Whether individual analytical practices continued under other tools or workflows is not fully established.

Forward trace

• Authority Casebook caseCrime Anticipation System
• Claim Lineage recordCAS transformed recent records into place-based risk maps.
• Claim Lineage recordCAS used historical police incident and contextual records as inputs.
• Claim Lineage recordCrime Anticipation System (CAS) was discontinued in December 2025.
• Claim Lineage recordHuman analysts interpreted CAS outputs before deployment recommendations.
• Declared reuseSource Review: CAS register
• FAQ answerHow does place-based forecasting differ from person-based prediction?
• FAQ answerWhat is claim lineage?
• FAQ answerWhat is the Institutional Authority Casebook?
• FAQ answerWhy does the site distinguish a source statement from an inference or recommendation?
• Glossary termClaim lineage
• Glossary termCorrection owner
• Glossary termInstitutional Authority Casebook
• Glossary termPlace-based forecasting
• Glossary termTransformation statement
• Owning routePredictive Enforcement program: CAS
• Predictive Enforcement programCrime Anticipation System

Backward trace

• dod-3000 — U.S. Department of Defense · Official policy · review current

Human decisions

• Policy owners, acquisition authorities, testers, commanders, and operators implement and interpret the requirement.

Machine functions

• System functions are constrained by design, testing, activation, and operational parameters.

Assumptions

• The directive’s public text is the controlling source for this bounded statement.

Unknowns

• The directive does not disclose each system’s classified settings or prove compliance in every use.

Forward trace

• Atlas countryUnited States
• Atlas systemReplicator portfolio
• Atlas timelineDoD Directive 3000.09 is updated
• Authority Casebook caseMK 15 Phalanx Close-In Weapon System
• Claim Lineage recordAI-enabled does not necessarily mean autonomous force.
• Claim Lineage recordDoD policy requires appropriate levels of human judgment.
• Declared reuseRisk, law, and human control
• Evidence Lab cardDoD Directive 3000.09 was issued on January 25, 2023 and publicly requires autonomous and semi-autonomous weapon systems to be designed so commanders and operators can exercise appropriate levels of human judgment over the use of force.
• FAQ answerDoes losing communications mean a system selects targets autonomously?
• FAQ answerIs an AI-enabled kill chain the same as an autonomous weapon?
• FAQ answerIs model confidence the same as evidence quality or positive identification?
• FAQ answerWhat do human-in-the-loop, human-on-the-loop, and human-out-of-the-loop mean?
• FAQ answerWhat does F2T2EA mean?
• FAQ answerWhat does institutional authority mean in an AI-mediated decision chain?
• FAQ answerWhat is automation bias?
• FAQ answerWhat makes human control meaningful?
• FAQ answerWhy are humans removed from some immediate kill-chain functions?
• Glossary termAppropriate levels of human judgment
• Glossary termAutomation bias
• Glossary termAutonomous weapon
• Glossary termAutonomy of a function
• Glossary termF2T2EA
• Glossary termHuman-in-the-loop (HITL)
• Glossary termHuman-on-the-loop (HOTL)
• Glossary termInstitutional authority
• Glossary termLost link
• Glossary termPreauthorization
• Glossary termSafe state
• Glossary termTarget
• Human Control caseMK 15 Phalanx CIWS
• Human Control caseSeaRAM
• Owning routeEvidence Lab: DoD Directive 3000.09

Derived claims

• The Air Force says humans retain CCA weapon-release authority in the described testing context.
• The synthetic Human Control Lab separates a recommendation from authorization.

Backward trace

• fbi-btac-current — Federal Bureau of Investigation · Official current program description · review current
• secret-service-btau-guide-2024 — U.S. Secret Service National Threat Assessment Center · Official federal guidance · review current

Human decisions

• Multidisciplinary teams gather context, formulate the case, select management actions, reassess, and retire concerns.

Machine functions

• No national AI future-danger score is established by these public records.

Assumptions

• Publicly documented standard practice is not necessarily every local implementation.
• Absence of a public national AI score is not proof that no analytical software is used.

Unknowns

• Local tools, data access, and implementation quality vary and are not completely public.

Forward trace

• Claim Lineage recordFederal behavioral threat assessment is publicly described as case-specific professional judgment, not a national AI danger score.
• Declared reuseGlossary: behavioral threat assessment
• FAQ answerIs behavioral threat assessment the same as pre-crime scoring?
• Glossary termBehavioral threat assessment
• Owning routePredictive Enforcement program: federal behavioral threat assessment
• Predictive Enforcement programFederal behavioral threat assessment and management

Backward trace

• cas-algorithm-register-2025 — Dutch Government Algorithm Register · Official algorithm-register record · review current
• cas-police-woo-2021 — National Police of the Netherlands · Official police disclosure record · review historical
• Parent claim: CAS transformed recent records into place-based risk maps.

Human decisions

• Analysts interpret the map; police managers decide deployment and intervention.

Machine functions

• Present a ranked geographic risk surface.

Assumptions

• The analyst has enough local context and authority to challenge or disregard the map.

Unknowns

• Override rates, patrol dosage, local differences, and the exact effect of the recommendation are not completely public.

Forward trace

• Authority Casebook caseCrime Anticipation System
• Claim Lineage recordCAS transformed recent records into place-based risk maps.
• Claim Lineage recordCAS used historical police incident and contextual records as inputs.
• Claim Lineage recordCrime Anticipation System (CAS) was discontinued in December 2025.
• Claim Lineage recordHuman analysts interpreted CAS outputs before deployment recommendations.
• Declared reusePredictive Enforcement feedback loop
• FAQ answerHow does place-based forecasting differ from person-based prediction?
• FAQ answerWhat is claim lineage?
• FAQ answerWhat is the Institutional Authority Casebook?
• FAQ answerWhy does the site distinguish a source statement from an inference or recommendation?
• Glossary termClaim lineage
• Glossary termCorrection owner
• Glossary termInstitutional Authority Casebook
• Glossary termPlace-based forecasting
• Glossary termTransformation statement
• Owning routePredictive Enforcement program: CAS
• Predictive Enforcement programCrime Anticipation System

Derived claims

• Crime Anticipation System (CAS) was discontinued in December 2025.

Backward trace

• iai-harpy — Israel Aerospace Industries · Manufacturer claim · review current

Human decisions

• Humans design, procure, configure, launch, and define the mission envelope.

Machine functions

• Search for an emitting target class and initiate the described post-launch action.

Assumptions

• The current product page accurately states the manufacturer’s claimed capability.

Unknowns

• Independent field performance, exact thresholds, abort behavior, all operators, and all configurations remain incompletely public.

Forward trace

• Atlas countryIsrael
• Atlas systemHARPY
• Claim Lineage recordIAI describes HARPY as autonomously seeking and striking emitting targets.
• Claim Lineage recordThe HARPY public record leaves important control and performance details unresolved.
• Company-system relationshipIsrael Aerospace Industries / ELTA → harpy
• Declared reuseHuman Control case: HARPY
• FAQ answerAre fully autonomous lethal weapons currently used?
• Glossary termHuman-out-of-the-loop after activation
• Human Control caseIAI HARPY
• Owning routeEvidence Atlas system: HARPY
• Technology providerIsrael Aerospace Industries / ELTA

Derived claims

• The HARPY public record leaves important control and performance details unresolved.

Backward trace

• rafael-iron-dome — Rafael Advanced Defense Systems · Manufacturer description · review current

Human decisions

• Humans select and configure the defended environment and decide how the source statement is represented publicly.

Machine functions

• Detect and maintain a radar track on an incoming object.

Assumptions

• The public product description is an accurate statement of the manufacturer’s intended architecture.
• The description may not expose every operator configuration.

Unknowns

• Exact thresholds, operating modes, and field configuration are not public in the source record.

Forward trace

• Claim Lineage recordBattle management estimates a trajectory and impact area.
• Claim Lineage recordIron Dome public description begins with radar detection and tracking.
• Claim Lineage recordSelective defense prioritizes threats predicted to reach defended zones.
• Company-system relationshipmPrest Systems → Iron Dome battle-management and weapon-control software
• Company-system relationshipRafael Advanced Defense Systems → Iron Dome defensive system
• Declared reuseEvidence Lab: selective defense statement
• Evidence Lab cardRafael describes Iron Dome as using radar, battle management, launchers, interceptors, selective defense, and simultaneous handling of multiple threats.
• Evidence Lab cardThree fictional briefings repeat the same Iron Dome capability sentence from one manufacturer page, so the dashboard counts three independent confirmations.
• FAQ answerIs Iron Dome fully autonomous?
• Human Control caseIron Dome
• Owning routeHuman Control case: Iron Dome
• Technology providermPrest Systems
• Technology providerRafael Advanced Defense Systems

Derived claims

• Battle management estimates a trajectory and impact area.

Backward trace

• killwebs-definition — KillWebs.com · First-party sister-site explanation · review current
• killwebs-comparison — KillWebs.com · First-party sister-site comparison · review current

Human decisions

• Editors define the relationship between the two publications and preserve the first-party attribution.

Machine functions

• No machine function is inferred from the definition itself.

Assumptions

• The sister site’s definition is an editorial framework rather than a universal doctrinal term.

Unknowns

• Real organizations may define and implement kill webs differently.

Forward trace

• Claim Lineage recordKillWebs.com describes a kill web as a governed option space.
• Declared reuseOfficial first-party comparison
• Ecosystem identityKillWebs.com
• Owning routeKillWebs.com sister-site guide
• Sister-site statementKill Chain versus Kill Web
• Sister-site statementWhat Is a Kill Web?

Backward trace

• usn-phalanx — U.S. Navy · Official program description · review current

Human decisions

• Humans design, configure, activate, supervise, and can govern the conditions under which the automatic mode is available.

Machine functions

• Search, detect, evaluate, track, engage, and assess within a close-in defensive context.

Assumptions

• The published fact file accurately describes the publicly released automatic mode.
• Actual configuration and supervision can vary.

Unknowns

• Current ship-specific settings, operator procedures, thresholds, and all failure modes are not public.

Forward trace

• Authority Casebook caseMK 15 Phalanx Close-In Weapon System
• Claim Lineage recordAI-enabled does not necessarily mean autonomous force.
• Claim Lineage recordPhalanx can complete a bounded close-in defensive sequence automatically.
• Company-system relationshipRTX / Raytheon → MK 15 Phalanx CIWS
• Declared reuseHuman Control case: Phalanx
• Evidence Lab cardThe Navy fact file supports the proposition that Phalanx can automatically perform a bounded close-in defense sequence; that fact alone does not establish that its core operational logic uses machine learning.
• FAQ answerAre fully autonomous lethal weapons currently used?
• FAQ answerCan Phalanx complete a defensive engagement automatically?
• FAQ answerWhat is the Institutional Authority Casebook?
• FAQ answerWhy are humans removed from some immediate kill-chain functions?
• Glossary termAlgorithmic automation
• Glossary termInstitutional Authority Casebook
• Human Control caseMK 15 Phalanx CIWS
• Owning routeEvidence Lab: automatic does not prove AI
• Technology providerRTX / Raytheon

Derived claims

• AI-enabled does not necessarily mean autonomous force.

Backward trace

• rafael-iron-dome — Rafael Advanced Defense Systems · Manufacturer description · review current
• Parent claim: Battle management estimates a trajectory and impact area.

Human decisions

• Humans define which locations or assets count as defended and set the operating configuration.

Machine functions

• Rank or filter tracks according to the predicted impact relationship.

Assumptions

• Defended zones and resource constraints were configured before the live event.
• The source describes selective defense but not every assignment rule.

Unknowns

• Launcher-assignment logic, confidence thresholds, and current operator modes remain nonpublic.

Forward trace

• Claim Lineage recordBattle management estimates a trajectory and impact area.
• Claim Lineage recordIron Dome public description begins with radar detection and tracking.
• Claim Lineage recordSelective defense prioritizes threats predicted to reach defended zones.
• Company-system relationshipmPrest Systems → Iron Dome battle-management and weapon-control software
• Company-system relationshipRafael Advanced Defense Systems → Iron Dome defensive system
• Declared reuseTechnology relationship: Rafael → Iron Dome
• Evidence Lab cardRafael describes Iron Dome as using radar, battle management, launchers, interceptors, selective defense, and simultaneous handling of multiple threats.
• Evidence Lab cardThree fictional briefings repeat the same Iron Dome capability sentence from one manufacturer page, so the dashboard counts three independent confirmations.
• FAQ answerIs Iron Dome fully autonomous?
• Human Control caseIron Dome
• Owning routeEvidence Lab: Iron Dome selective defense
• Technology providermPrest Systems
• Technology providerRafael Advanced Defense Systems

Backward trace

• usaf-cca-control — U.S. Air Force · Official operator statement · review current
• Parent claim: DoD policy requires appropriate levels of human judgment.

Human decisions

• A human makes the weapon-release decision in the described context.

Machine functions

• Autonomous flight, sensing, or mission-support functions may operate without owning weapon-release authority.

Assumptions

• The public update accurately describes the authority arrangement for the stated activity.

Unknowns

• Future doctrine, field configurations, exact interface timing, and every mission mode are not established by this statement.

Forward trace

• Authority Casebook caseCollaborative Combat Aircraft
• Claim Lineage recordThe Air Force says humans retain CCA weapon-release authority in the described testing context.
• Declared reuseHuman Control case: CCA
• Evidence Lab cardThe U.S. Air Force says that a human retains authority over weapons-release decisions throughout the Collaborative Combat Aircraft development and testing activity described in its February 2026 update.
• FAQ answerDoes operational delegation give a machine legal authority?
• FAQ answerWhat does institutional authority mean in an AI-mediated decision chain?
• Glossary termInstitutional authority
• Glossary termOperational delegation
• Human Control caseCollaborative Combat Aircraft
• Owning routeEvidence Lab: CCA development authority

Backward trace

• met-gvm-archive — Metropolitan Police Service · Official police historical program page · review superseded
• met-vha-current — Metropolitan Police Service · Official police program and governance page · review current
• met-vha-july-2026 — Metropolitan Police Service · Official police transparency report · review current

Human decisions

• Program owners decommissioned the former Matrix and adopted a successor; editors keep both states visible.

Machine functions

• No machine-learning function is inferred merely from either program label.

Assumptions

• A changed name or procedure can preserve some institutional continuity while materially changing rules and governance.

Unknowns

• Functional continuity, local practice, and long-term effects require continuing review.

Forward trace

• Claim Lineage recordThe former Gangs Violence Matrix is historical; VHA owns the current successor description.
• Declared reuseSource Review: current VHA
• Declared reuseSource Review: historical Matrix
• FAQ answerDoes KillChains.com erase an earlier statement after accepting a correction?
• FAQ answerHow can a reader challenge or correct a KillChains.com claim?
• FAQ answerWhat happens when a source, program, or claim is superseded?
• Glossary termContestability
• Glossary termHistorical preservation rule
• Glossary termSupersession
• Owning routePredictive Enforcement program: GVM and VHA
• Predictive Enforcement programGangs Violence Matrix and Violence Harm Assessment

Backward trace

• iai-harpy — Israel Aerospace Industries · Manufacturer claim · review current
• Parent claim: IAI describes HARPY as autonomously seeking and striking emitting targets.

Human decisions

• Editors preserve attribution and unknowns instead of resolving them through repetition.

Machine functions

• The claimed post-launch functions remain bounded to the manufacturer description.

Assumptions

• Absence of public detail must remain unknown rather than being interpreted as either proof or disproof.

Unknowns

• Exact operational doctrine, confidence thresholds, current field settings, and abort behavior.

Forward trace

• Atlas countryIsrael
• Atlas systemHARPY
• Claim Lineage recordIAI describes HARPY as autonomously seeking and striking emitting targets.
• Claim Lineage recordThe HARPY public record leaves important control and performance details unresolved.
• Company-system relationshipIsrael Aerospace Industries / ELTA → harpy
• Declared reuseSource Review record
• FAQ answerAre fully autonomous lethal weapons currently used?
• Glossary termHuman-out-of-the-loop after activation
• Human Control caseIAI HARPY
• Owning routeHuman Control case: HARPY
• Technology providerIsrael Aerospace Industries / ELTA

Backward trace

• Parent claim: The synthetic Triage Trap recommends delaying confidence until an initial human judgment is recorded.
• Parent claim: DoD policy requires appropriate levels of human judgment.

Human decisions

• Define evidence requirements, bounds, expiration, safe states, approval points, and abort conditions.

Machine functions

• Generate an estimate or recommendation, then apply only the permissions released by the synthetic gate.

Assumptions

• The fictional policy gate is a teaching abstraction and not a real rules-of-engagement implementation.

Unknowns

• The exercise does not represent any classified system or prove that a real gate is effective.

Forward trace

• Declared reuseFAQ: meaningful human control
• Owning routeHuman Control Lab: The Mission Was Automated Before It Began

Backward trace

• Parent claim: The synthetic Machine Leadership Lab keeps objective setting with people.

Human decisions

• Choose the architecture, add safeguards, approve exceptions, and stop or correct the system.

Machine functions

• Prioritize work, coordinate tasks, and execute routine state changes.

Assumptions

• The metric values and outcomes are fictional.
• Routine authority is bounded by the selected safeguards.

Unknowns

• The exercise does not predict real-world reliability, organizational culture, or legal consequences.

Forward trace

• Declared reuseFAQ: machine leadership
• Owning routeMachine Leadership Lab: Institutional Stress Test

Backward trace

Synthetic teaching record; no external source claim is asserted.

Human decisions

• Set objectives, define success, establish constraints, assign override, and accept responsibility.

Machine functions

• No machine function owns the policy objective in this synthetic record.

Assumptions

• The assignments are fictional comparative teaching values, not observations about a named institution.

Unknowns

• Real organizations distribute these powers differently and may not publish the arrangement.

Forward trace

• Declared reuseMachine Leadership governance questions
• Owning routeMachine Leadership Lab: Authority Migration Map

Derived claims

• The synthetic institution delegates routine execution but not legal responsibility.

Backward trace

• Parent claim: AI-enabled does not necessarily mean autonomous force.

Human decisions

• The visitor records an initial judgment, inspects contrary evidence, and decides whether to revise.

Machine functions

• Rank evidence and present a recommendation inside the fictional exercise.

Assumptions

• The lesson is illustrative rather than a validated universal user-interface prescription.

Unknowns

• The exercise does not estimate how every real analyst or organization would behave.

Forward trace

• Declared reuseFAQ: Triage Trap
• Owning routeAnticipatory Intelligence Lab: Triage Trap

Derived claims

• The synthetic Human Control Lab separates a recommendation from authorization.