{"$schema":"https://policywindow.org/wiki/accountability-completeness.json","name":"Policy Window accountability-completeness measure","description":"Per harm (topic): which accountability lever categories act on it (binding law / soft-voluntary / technical standard / enforcement-litigation), which are absent, and whether any rigorous evidence shows the governance works. Surfaces 'named but toothless' harms (a binding rule with no catalogued enforcement) and the uniform absence of efficacy evidence. Measures the CATALOG'S RECORD, not a verdict on real-world protection.","docs":"https://policywindow.org/wiki/accountability-completeness","leverCategories":{"bindingLaw":"binding_regulation / executive_order / international_treaty / delegated_act / implementing_regulation","softLaw":"voluntary_code / policy_statement / resolution","standard":"technical_standard","enforcement":"a catalogued enforcement / litigation case tagged to the topic"},"summary":{"topics":23,"meanLeverTypes":2.783,"namedButToothless":13,"softOnly":0,"noEfficacyEvidence":23,"ungoverned":0,"bindingButUnproven":23},"byLever":[{"category":"bindingLaw","topicsWithLever":23,"share":1},{"category":"softLaw","topicsWithLever":22,"share":0.9565},{"category":"standard","topicsWithLever":9,"share":0.3913},{"category":"enforcement","topicsWithLever":10,"share":0.4348}],"topics":[{"code":"sovereign_ai","label":"Sovereign AI Doctrine","bindingLaw":5,"softLaw":0,"standard":0,"enforcement":0,"leverTypesPresent":1,"efficacyStatus":"absent","flags":["named_but_toothless","no_efficacy_evidence"]},{"code":"ai_worker_displacement","label":"AI-Driven Worker Displacement","bindingLaw":3,"softLaw":6,"standard":0,"enforcement":0,"leverTypesPresent":2,"efficacyStatus":"thin","flags":["named_but_toothless","no_efficacy_evidence"]},{"code":"compute_export_controls","label":"Compute + Model-Weight Export Controls","bindingLaw":1,"softLaw":3,"standard":0,"enforcement":0,"leverTypesPresent":2,"efficacyStatus":"absent","flags":["named_but_toothless","no_efficacy_evidence"]},{"code":"compute_reporting","label":"Compute-Threshold Reporting","bindingLaw":7,"softLaw":9,"standard":0,"enforcement":0,"leverTypesPresent":2,"efficacyStatus":"absent","flags":["named_but_toothless","no_efficacy_evidence"]},{"code":"education","label":"AI in Education","bindingLaw":3,"softLaw":2,"standard":0,"enforcement":0,"leverTypesPresent":2,"efficacyStatus":"thin","flags":["named_but_toothless","no_efficacy_evidence"]},{"code":"international_coordination","label":"International Coordination","bindingLaw":3,"softLaw":13,"standard":0,"enforcement":0,"leverTypesPresent":2,"efficacyStatus":"absent","flags":["named_but_toothless","no_efficacy_evidence"]},{"code":"national_security_carveouts","label":"National Security Carveouts in AI Regulation","bindingLaw":8,"softLaw":4,"standard":0,"enforcement":0,"leverTypesPresent":2,"efficacyStatus":"absent","flags":["named_but_toothless","no_efficacy_evidence"]},{"code":"open_weight_release","label":"Open-Weight Frontier Release","bindingLaw":5,"softLaw":7,"standard":0,"enforcement":0,"leverTypesPresent":2,"efficacyStatus":"absent","flags":["named_but_toothless","no_efficacy_evidence"]},{"code":"tech_sovereignty","label":"Technological Sovereignty","bindingLaw":5,"softLaw":6,"standard":0,"enforcement":0,"leverTypesPresent":2,"efficacyStatus":"absent","flags":["named_but_toothless","no_efficacy_evidence"]},{"code":"agentic_systems_governance","label":"Agentic AI Governance","bindingLaw":9,"softLaw":7,"standard":2,"enforcement":0,"leverTypesPresent":3,"efficacyStatus":"absent","flags":["named_but_toothless","no_efficacy_evidence"]},{"code":"catastrophic_risk","label":"Catastrophic & Existential Risk","bindingLaw":7,"softLaw":13,"standard":2,"enforcement":0,"leverTypesPresent":3,"efficacyStatus":"absent","flags":["named_but_toothless","no_efficacy_evidence"]},{"code":"environmental_impact_of_training","label":"Environmental Impact of AI Training","bindingLaw":4,"softLaw":7,"standard":1,"enforcement":0,"leverTypesPresent":3,"efficacyStatus":"absent","flags":["named_but_toothless","no_efficacy_evidence"]},{"code":"synthetic_content_provenance","label":"Synthetic Content Provenance","bindingLaw":7,"softLaw":8,"standard":2,"enforcement":0,"leverTypesPresent":3,"efficacyStatus":"thin","flags":["named_but_toothless","no_efficacy_evidence"]},{"code":"biometric_id","label":"Biometric Identification","bindingLaw":6,"softLaw":2,"standard":0,"enforcement":3,"leverTypesPresent":3,"efficacyStatus":"thin","flags":["no_efficacy_evidence"]},{"code":"criminal_justice","label":"AI in Criminal Justice","bindingLaw":4,"softLaw":2,"standard":0,"enforcement":1,"leverTypesPresent":3,"efficacyStatus":"thin","flags":["no_efficacy_evidence"]},{"code":"development_rights_framing","label":"Development-Rights Framings","bindingLaw":5,"softLaw":6,"standard":0,"enforcement":1,"leverTypesPresent":3,"efficacyStatus":"absent","flags":["no_efficacy_evidence"]},{"code":"employment","label":"AI in Employment","bindingLaw":5,"softLaw":3,"standard":0,"enforcement":2,"leverTypesPresent":3,"efficacyStatus":"absent","flags":["no_efficacy_evidence"]},{"code":"healthcare","label":"AI in Healthcare","bindingLaw":4,"softLaw":3,"standard":0,"enforcement":1,"leverTypesPresent":3,"efficacyStatus":"absent","flags":["no_efficacy_evidence"]},{"code":"deepfakes","label":"Deepfakes / Synthetic Content","bindingLaw":8,"softLaw":4,"standard":2,"enforcement":3,"leverTypesPresent":4,"efficacyStatus":"thin","flags":["no_efficacy_evidence"]},{"code":"foundation_models","label":"Foundation Models / GPAI","bindingLaw":13,"softLaw":19,"standard":2,"enforcement":5,"leverTypesPresent":4,"efficacyStatus":"absent","flags":["no_efficacy_evidence"]},{"code":"redress","label":"Individual Redress","bindingLaw":15,"softLaw":10,"standard":2,"enforcement":11,"leverTypesPresent":4,"efficacyStatus":"absent","flags":["no_efficacy_evidence"]},{"code":"training_data","label":"Training-Data Rights","bindingLaw":10,"softLaw":8,"standard":2,"enforcement":5,"leverTypesPresent":4,"efficacyStatus":"absent","flags":["no_efficacy_evidence"]},{"code":"transparency","label":"Transparency Obligations","bindingLaw":18,"softLaw":21,"standard":2,"enforcement":9,"leverTypesPresent":4,"efficacyStatus":"absent","flags":["no_efficacy_evidence"]}],"caveat":"This measures the CATALOG'S RECORD of accountability levers per harm, not a verdict on real-world protection. 'No enforcement' means none is catalogued (the catalog tracks a finite set of enforcement cases — absence may reflect catalog coverage rather than reality); 'efficacy absent/thin' means no rigorous evaluation shows the governance works (absence of evidence, not evidence of failure). A binding rule with no catalogued enforcement and no efficacy evidence is flagged as paper protection to prompt scrutiny, not as a proof of failure. Lever counts are by an instrument's legal KIND regardless of its lifecycle STATUS — a counted 'binding rule' may be proposed, partially rescinded, or adopted-not-yet-in-force (a few harms are backed only by a rescinded executive order or an un-enacted bill), so 'a binding rule reaches every harm' means a binding instrument is catalogued, not that a live statute is in force.","relatedMeasures":{"coverageEquity":"https://policywindow.org/wiki/coverage-equity","evidenceMap":"https://policywindow.org/wiki/evidence"},"license":"CC BY 4.0","attribution":"Policy Window — policywindow.org"}