Composite Licensing as Authority Intersection
by Nick Clark | Published April 25, 2026
When a skill artifact composes content from multiple authorities — training-data licensors, fine-tuning licensors, deployment licensors, downstream-use licensors — the licensing emerges from the intersection of credentialed authorities. This supports rights-grade generative AI without per-licensor renegotiation, an architectural answer to the structural rights problem the generative AI ecosystem currently faces.
What Composite Licensing Specifies
Each adaptation artifact carries credentialed metadata identifying the licensing authorities relevant to its use: who licensed the training data, who licensed the fine-tuning process, who licensed the deployment context, who has standing to grant or deny downstream use. The licensing decision for any specific use is the intersection of all relevant authorities' policies.
When the consumer deploys the artifact, the admissibility gate evaluates the licensing intersection: do all relevant authorities admit this specific use under this consumer's policy. The result is a credentialed observation — either the artifact is admissible for this use, or specific authority refusals identify the licensing gap.
Why Per-Authority Renegotiation Doesn't Scale
Generative AI's licensing problem is structurally a multi-authority problem. The training data has hundreds of source authorities. The fine-tuning process has its own authority. The base model has its authority. The deployment context has its policy. The downstream use has its requirements. Per-authority renegotiation does not scale to the dimensionality of the problem.
The current pattern — opt-in licensing, opt-out registries, blanket fair-use claims, content-creator lawsuits — is symptom rather than architectural solution. Each pattern attempts to compress the multi-authority problem into a single-authority decision, and each fails as the dimensionality grows.
How Authority Intersection Operates
Each authority signs a policy describing what uses they admit. The artifact's metadata identifies the relevant authorities. The composite admissibility evaluation evaluates each authority's policy against the proposed use; the use is admissible only if all relevant authorities admit.
Authority intersection naturally handles complex cases. A training-data authority that admits research use but not commercial use produces an artifact that is research-admissible but not commercial-admissible. A deployment authority that requires specific compensation for use produces an admissibility result that includes the compensation routing. Cross-authority resolution handles disputes through the same governance framework that handles other multi-authority disputes.
What This Enables for Rights-Grade Generative AI
The training-data lawsuits currently active against major AI vendors (NYT vs OpenAI, the music labels' Anthropic suits, image-generator class actions) all share a structural pattern: rights authorities want compensation and use control that the current architecture cannot provide. The architecture has no concept of authority intersection.
Composite licensing through authority intersection provides the structural answer. Authoring authorities sign artifacts with the relevant authority chain; consumers admit through composite admissibility; compensation routes through the credentialed authority chain. The rights-grade generative AI that the lawsuits are converging toward becomes structurally tractable. The patent positions the primitive at the layer the rights problem actually requires.
