Content Provenance for Video and Broadcast Media

Video Is the Hardest Problem

Video presents the most challenging content provenance problem in the media ecosystem. A single broadcast-quality clip passes through ingest systems, editing suites, graphics packages, compression pipelines, CDN delivery, and social clips — each step potentially introducing format changes that can disrupt provenance metadata.

Text and images are relatively tractable. Video, with its complex container formats, multiple streams, and multi-step post-production workflows, requires a different approach.

How C2PA Handles Video

The C2PA specification supports video provenance through manifest embedding in container formats (MP4, MXF, MOV) and through sidecar manifest files when embedding is not possible. Key assertions for video include:

• Creative work: title, description, and copyright
• Actions: record of every edit, transcode, and delivery step
• AI generative: flag for synthetic or AI-assisted content
• Ingredients: source clips and their own provenance records

Broadcast Workflows

Major broadcast organizations — including BBC, CBC/Radio-Canada, and Reuters — are actively implementing C2PA in their production workflows. The challenge is integrating provenance at ingest, before content enters post-production pipelines that have no native C2PA support.

Limbo’s API-first architecture is designed for this integration point: sign and watermark content at ingest, then let it flow through existing workflows. The provenance record travels with the content; downstream steps append their own entries without requiring changes to editing or delivery tools.