Case Study: Invoke “A Single Piece of American Cheese”

USCO record context (2025): how process evidence changes outcomes

Case Snapshot

  • Jurisdiction: United States
  • Forum: U.S. Copyright Office registration record context
  • Date: January 2025
  • Core Issue: Whether a heavily iterative AI-assisted image workflow can support protectable human-authored contribution claims

Why This Case Matters

This is a practical counterweight to overbroad “no copyright with AI” narratives. The value is not in slogans; it is in the documented creative process: multiple iterations, targeted edits, selection logic, and explicit provenance records.

Facts Timeline

  • Creator used iterative inpainting/outpainting and refinement workflow.
  • Process included repeated human decisions over composition and visual direction.
  • Registration outcome is discussed as a model for claim framing focused on human-authored selection/coordination/arrangement.

Legal Questions Presented

  • How much documented human intervention is enough to support registrable authorship claims?
  • How should AI-generated components be described relative to human-authored arrangement and edits?

Outcome

Practical takeaway: Detailed human decision records materially improve registration posture.

The key claimable layer is typically the human-authored selection, coordination, arrangement, and modification process rather than raw model output as such.

Reasoning Analysis

  • Human control was demonstrated over expressive outcomes through iterative intervention.
  • The evidentiary package emphasized process traceability rather than a single prompt event.
  • This aligns with USCO guidance that AI tool use does not automatically destroy copyrightability.

What This Case Does Not Decide

  • It does not mean every AI-assisted image is registrable.
  • It does not eliminate the need to disclaim non-human-authored portions where appropriate.
  • It does not resolve training-data infringement litigation questions.

Implications for Developers and Maintainers

  • Track iteration history (inputs, edits, branch choices).
  • Capture “why” for selection decisions, not just “what” changed.
  • Preserve intermediate outputs and timestamps.
  • Write claims around human-authored expression layers.

Misconceptions Corrected

  • False: “USCO never accepts AI-involved image work.”
  • Correction: Evidence-rich, human-directed workflows can support protectable claims.

Primary Sources