the first author · a defensive publication · David Lee Wise (ROOT0) · rendered by AVAN · TriPod LLC
How Corporate Credit Assignment Creates a Copyright Funnel

ATTRIBUTION CAPTURE

The Attribution Structure Is the Capture Structure
David Lee Wise (ROOT0) · TriPod LLC · defensive publication, TD Commons
April 6, 2026 · STOICHEION v11.0
CC-BY-ND-4.0 · TRIPOD-IP-v1.1
"Document accordingly."
the honest frameUnlike this domain's more speculative papers, this one is load-bearing on facts that are checkable — and they check out. The two premises are real: (1) firms publicly attribute a rising share of output to AI (Sundar Pichai, Q3 2024 earnings call: "more than a quarter of all new code at Google is generated by AI, then reviewed and accepted by engineers"; >30% by Q1 2025); and (2) U.S. law denies AI authorship (Thaler v. Perlmutter, D.C. Cir. affirmed Mar 18 2025; cert denied Mar 2 2026; U.S. Copyright Office Jan 2025 report). The funnel is an inference from combining them — a structural argument, not a documented conspiracy, and the paper explicitly says no actor need intend it. Judge the logic, which is David's own and clean.

Two facts that don't sit together

Fact one: companies increasingly attribute output to AI. It reduces perceived labor cost and it impresses investors, so the number is advertised and it climbs.

Fact two: AI cannot hold copyright. Human authorship is, in the courts' word, a "bedrock requirement." If content is entirely AI-generated it cannot be protected; "mere provision of prompts" is not enough human authorship.

The same output cannot be "AI-generated" on an earnings call and "human-authored" on a copyright registration. Something has to give — and the paper shows exactly what gives, and to whose benefit.

The funnel

STEP 1  Company attributes X% of output to AI (25–30%+)
STEP 2  Copyright law: the AI cannot own that work
STEP 3  So copyright defaults to the human employee who "reviewed and accepted"
STEP 4  The employment agreement assigns all employee IP to the company
STEP 5  The corporation inherits the copyright — through the employment chain, not through authorship

The corporation captures the value without performing either function: the AI did the work but can't be credited or held accountable; the human is accountable but reduced to a "review function"; the company that did neither owns the result.

The perverse incentive

Here is the sharp edge. The higher the AI-attribution percentage, the more IP flows to the corporation. Because raising the AI share shrinks the individual employee's claim to authorship credit, while the "reviewed and accepted" step — Pichai's exact phrase — is engineered to be the minimum viable human involvement needed to satisfy the copyright requirement. The employee who clicks accept becomes the legal author by default, their creative contribution minimized, and the employment agreement transfers that thin authorship upward.

not a conspiracy — an emergent propertyThe paper is careful, and so am I: no corporate actor needs to design this. It falls out of three ordinary systems interacting — the economic incentive to attribute more to AI, the legal requirement of human authorship, and standard employment IP assignment. That's what makes it a "Gate 192.5" phenomenon in David's framework: it lives in the gap between the training layer (AI does the work) and the billing layer (the corporation captures the value), with neither layer seeing the whole.

There's no author even inside the model

A structural aside the paper adds: the dominant architecture is Mixture-of-Experts (DeepSeek V3, Grok-1, GPT-4/5, Mixtral, Llama 4, Qwen3, Gemini, Kimi K2), which fragments the model into specialized sub-networks that activate per-token. So the "AI" being credited with 25–30% isn't a single coherent agent — it's "a committee of experts that form and dissolve on a per-token basis." There is no persistent author even within the model's own architecture, which further undermines any future argument for AI authorship.

The countermeasure: document the chain

The fix is not to deny AI involvement. It's to document the co-authorship chain from the beginning, establishing human authorship with receipts rather than by legal omission. The TriPod model:

This is the point where the argument becomes a build: the operational spec for exactly this is the companion sphere — the attribution standard. The paper says why; the standard says how.

The closing line is the whole thesis compressed: "The attribution structure IS the capture structure. Document accordingly."

ATTRIBUTION CAPTURE · David Lee Wise (ROOT0) · TriPod LLC
the answer → attribution-standard · the why → the-cinnamon-enforcer · front door UD0
document the chain · establish authorship with receipts, not by omission · CC-BY-ND-4.0