Adobe Stock as Governance for AI-Generated News Images

• In Event: Student Forum Roundtables Session
  In Refereed Roundtable (90 minute): Roundtable 8: AI and Digital Governance

Abstract

This paper analyzes Adobe Stock as an upstream governance infrastructure for AI-generated conflict imagery, asking how provenance standards and platform rules shape the circulation of synthetic visuals before they reach newsrooms or publics. Rather than treating deceptive or ambiguous AI images only as downstream editorial failures, the study examines how stock-platform governance organizes what becomes available, visible, and licensable in the first place. The paper uses a two-layer qualitative case design: (1) a purposive corpus of 21 documents from Adobe, CAI, and C2PA, and (2) a dated interface audit of Adobe Stock search results for the keyword “war,” coding the first 20 results for AI labels, provenance cues, asset type, category, and interface placement. Findings show three patterns. First, Adobe/CAI/C2PA frame Content Credentials as provenance infrastructure rather than truth verification, which limits what disclosure can guarantee in news-like contexts. Second, the policy corpus distributes responsibility across creators, platforms, publishers, and users through a progressive disclosure model, making interface visibility central to whether transparency works in practice. Third, Adobe Stock supplements provenance with preventive platform governance indicating that provenance alone is insufficient to manage “newsy” synthetic imagery. The paper argues that AI-generated conflict imagery should be studied not only as content, but as a platform-governed sociotechnical supply chain in which labeling, classification, and interface design shape the boundary between illustration and documentation.

Author

• Emilia Edwards, University of Texas-Austin