SUBROGATION FRAMEWORK
After GIPL pays a claim, who bears ultimate responsibility? The framework architecture for upstream recovery pathways from AI model vendors, training data suppliers, platform operators, and infrastructure providers.
The architectural premise
Insurance subrogation allows the insurer, after paying a claim, to step into the insured's shoes and pursue recovery against the party ultimately responsible. In AI-IP claims, the upstream chain is unusually complex: the AI vendor, the training data aggregator, the cloud infrastructure provider, and the platform operator may all have contributed to the loss.
The four upstream pathways
Pathway 1 — AI model vendor (the entity operating the model that generated the infringing output). Pathway 2 — Training data supplier (the entity that assembled or licensed the training corpus). Pathway 3 — Platform operator (the entity that deployed the model in a user-facing product). Pathway 4 — Infrastructure provider (cloud/compute provider, potentially liable under contributory infringement theories in some jurisdictions).
Operational dependencies — what the framework supplies versus what it does not
The framework supplies: CIP-AI-Vendor field for identifying the model vendor, CIP-AI-Training-Data-Source for training data provenance, Output-Provenance Architecture for output chain documentation, and AI-IP risk-index factor-level scores for liability apportionment. The framework does not supply: legal opinions on subrogation viability, jurisdictional enforceability assessments, or quantum calculations.
Integration with framework features
Integrates with GIPL Architecture (v3.8) for claims trigger, AI-IP Risk Architecture (v3.7) for liability apportionment factors, Output-Provenance Architecture (v3.17) for chain-of-causation documentation, and the AI Generation field set (v3.24) for vendor and model identification.
Jurisdictional regime mapping
UK: Insurance Act 2015, Third Parties (Rights against Insurers) Act 2010. US: state-by-state subrogation law, made-whole doctrine variations. EU: national insurance law implementations, Rome I Regulation for cross-border disputes. Subrogation viability varies materially by jurisdiction.
Worked subrogation case material
Worked examples tracing subrogation pathways through the framework's metadata infrastructure: voice-cloning claim traced to AI vendor via CIP-AI-Model field, training data copyright claim traced to data supplier via CIP-AI-Training-Data-Source, and platform liability claim traced via Output-Provenance chain.
CIP Subrogation Framework v3.21 (Path B expansion v3.23), https://creativeip.org/subrogation