# VideoSEAL: Mitigating Evidence Misalignment in Agentic Long Video Understanding by Decoupling Answer Authority

arXiv: https://arxiv.org/abs/2605.12571 · HTML: https://mlsanity.com/paper/2605.12571/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- Placed in Multimodal and embodied agents, one of 16 directions in LLM agents and tool use.
- Builds on 44 of its 53 references that are in the map.
- First posted on arXiv 12 May 2026.

## Placed in

- LLM agents and tool use / [Multimodal and embodied agents](https://mlsanity.com/agents/multimodal-and-embodied-agents/)

## Builds on (most cited first, 30 of 44)

- [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://mlsanity.com/paper/2402.03300/) (2024)
- [ReAct: Synergizing Reasoning and Acting in Language Models](https://mlsanity.com/paper/2210.03629/) (2022)
- [Efficient Memory Management for Large Language Model Serving with PagedAttention](https://mlsanity.com/paper/2309.06180/) (2023)
- [Lost in the Middle: How Language Models Use Long Contexts](https://mlsanity.com/paper/2307.03172/) (2023)
- [LLaVA-OneVision: Easy Visual Task Transfer](https://mlsanity.com/paper/2408.03326/) (2024)
- [InternVL3.5: Advancing Open-Source Multimodal Models in Versatility, Reasoning, and Efficiency](https://mlsanity.com/paper/2508.18265/) (2025)
- [LongBench: A Bilingual, Multitask Benchmark for Long Context Understanding](https://mlsanity.com/paper/2308.14508/) (2023)
- [Video-MME: The First-Ever Comprehensive Evaluation Benchmark of Multi-modal LLMs in Video Analysis](https://mlsanity.com/paper/2405.21075/) (2024)
- [Video-LLaVA: Learning United Visual Representation by Alignment Before Projection](https://mlsanity.com/paper/2311.10122/) (2023)
- [HuggingGPT: Solving AI Tasks with ChatGPT and its Friends in Hugging Face](https://mlsanity.com/paper/2303.17580/) (2023)
- [Dense-Captioning Events in Videos](https://mlsanity.com/paper/1705.00754/) (2017)
- [VideoLLaMA 2: Advancing Spatial-Temporal Modeling and Audio Understanding in Video-LLMs](https://mlsanity.com/paper/2406.07476/) (2024)
- [LongVideoBench: A Benchmark for Long-context Interleaved Video-Language Understanding](https://mlsanity.com/paper/2407.15754/) (2024)
- [MovieChat: From Dense Token to Sparse Memory for Long Video Understanding](https://mlsanity.com/paper/2307.16449/) (2023)
- [Long Context Transfer from Language to Vision](https://mlsanity.com/paper/2406.16852/) (2024)
- [From Generation to Judgment: Opportunities and Challenges of LLM-as-a-judge](https://mlsanity.com/paper/2411.16594/) (2024)
- [Video-R1: Reinforcing Video Reasoning in MLLMs](https://mlsanity.com/paper/2503.21776/) (2025)
- [VideoAgent: Long-form Video Understanding with Large Language Model as Agent](https://mlsanity.com/paper/2403.10517/) (2024)
- [VideoTree: Adaptive Tree-based Video Representation for LLM Reasoning on Long Videos](https://mlsanity.com/paper/2405.19209/) (2024)
- [Defining and Characterizing Reward Hacking](https://mlsanity.com/paper/2209.13085/) (2022)
- [Self-Chained Image-Language Model for Video Localization and Question Answering](https://mlsanity.com/paper/2305.06988/) (2023)
- [Local-Global Video-Text Interactions for Temporal Grounding](https://mlsanity.com/paper/2004.07514/) (2020)
- [A Simple LLM Framework for Long-Range Video Question-Answering](https://mlsanity.com/paper/2312.17235/) (2023)
- [Can I Trust Your Answer? Visually Grounded Video Question Answering](https://mlsanity.com/paper/2309.01327/) (2023)
- [Reward Tampering Problems and Solutions in Reinforcement Learning: A Causal Influence Diagram Perspective](https://mlsanity.com/paper/1908.04734/) (2019)
- [Deep Video Discovery: Agentic Search with Tool Use for Long-form Video Understanding](https://mlsanity.com/paper/2505.18079/) (2025)
- [CG-Bench: Clue-grounded Question Answering Benchmark for Long Video Understanding](https://mlsanity.com/paper/2412.12075/) (2024)
- [VideoRAG: Retrieval-Augmented Generation with Extreme Long-Context Videos](https://mlsanity.com/paper/2502.01549/) (2025)
- [Ego-R1: Chain-of-Tool-Thought for Ultra-Long Egocentric Video Reasoning](https://mlsanity.com/paper/2506.13654/) (2025)
- [DrVideo: Document Retrieval Based Long Video Understanding](https://mlsanity.com/paper/2406.12846/) (2024)

## Built on by (most cited first, 6 of 6)

- [VideoHarness-RSI: Recursive Harness Self-Improvement for Long-Video Understanding with Frozen Vision-Language Models](https://mlsanity.com/paper/2608.24302/) (2026)
- [Accuracy and Cost Claims Do Not Survive Re-Execution in Agentic VideoQA](https://mlsanity.com/paper/2607.01469/) (2026)
- [Counterfactual Sensitivity Is Not Repairability: Auditing Replay Probes for Video Evidence](https://mlsanity.com/paper/2608.15685/) (2026)
- [Agentic Video Understanding: A Survey](https://mlsanity.com/paper/2609.31713/) (2026)
- [CASE: Cost-Aware Stopping for Efficient Long-Video Agents](https://mlsanity.com/paper/2610.05400/) (2026)
- [VAMR: Multi-Question Agentic Reasoning for Efficient Long-Form Video Understanding](https://mlsanity.com/paper/2610.11171/) (2026)

## Abstract (arXiv)

Long video question answering requires locating sparse, time-scattered visual evidence within highly redundant content. Although current MLLMs perform well on short videos, long videos introduce long-horizon search and verification, which often necessitates multi-turn, agentic interaction. We show that existing LVU agents can exhibit "evidence misalignment": they produce correct answers that are not supported by the retrieved or inspected evidence. To characterize this failure, we introduce two diagnostics (temporal groundedness and semantic groundedness) and use them to reveal two pressures that amplify misalignment: prompt pressure from shared-context saturation at inference time and reward pressure from outcome-only optimization during training. These findings point to a structural root cause: the coupled agent paradigm conflates long-horizon planning with answer authority. We therefore propose the decoupled planner-inspector framework, which separates planning from answer authority and gates final answering on pixel-level verification. Across four long-video benchmarks, our framework improves both answer accuracy and evidence alignment, achieving 55.1% on LVBench and 62.0% on LongVideoBench while producing interpretable search trajectories. Moreover, the decoupled architecture scales consistently with increased search budgets and supports plug-and-play upgrades of the MLLM backbone without retraining the planner. Code and models are available at https://github.com/Echochef/VideoSEAL.