# AI vs. Human Moderators: A Comparative Evaluation of Multimodal LLMs in Content Moderation for Brand Safety

arXiv: https://arxiv.org/abs/2508.05527 · HTML: https://mlsanity.com/paper/2508.05527/

## Key facts

- Cited by 8 papers in the map (data through 14 Oct 2026).
- Placed in MLLM jailbreaks and adversarial robustness, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 10 of its 23 references that are in the map.
- First posted on arXiv 7 Aug 2025.

## Placed in

- Multimodal LLMs / vision-language models / [MLLM jailbreaks and adversarial robustness](https://mlsanity.com/mllm/mllm-jailbreaks-and-adversarial-robustness/)

## Builds on (most cited first, 10 of 10)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [Chain-of-Thought Prompting Elicits Reasoning in Large Language Models](https://mlsanity.com/paper/2201.11903/) (2022)
- [Robust Speech Recognition via Large-Scale Weak Supervision](https://mlsanity.com/paper/2212.04356/) (2022)
- [Exploiting Cloze Questions for Few Shot Text Classification and Natural Language Inference](https://mlsanity.com/paper/2001.07676/) (2020)
- [Automated Hate Speech Detection and the Problem of Offensive Language](https://mlsanity.com/paper/1703.04009/) (2017)
- [YouTube-8M: A Large-Scale Video Classification Benchmark](https://mlsanity.com/paper/1609.08675/) (2016)
- [Large Language Models for Data Annotation and Synthesis: A Survey](https://mlsanity.com/paper/2402.13446/) (2024)
- [AnnoLLM: Making Large Language Models to Be Better Crowdsourced Annotators](https://mlsanity.com/paper/2303.16854/) (2023)
- [Content Moderation by LLM: From Accuracy to Legitimacy](https://mlsanity.com/paper/2409.03219/) (2024)
- [Beyond Toxic: Toxicity Detection Datasets are Not Enough for Brand Safety](https://mlsanity.com/paper/2303.15110/) (2023)

## Built on by (most cited first, 8 of 8)

- [LLM Performance Predictors: Learning When to Escalate in Hybrid Human-AI Moderation Systems](https://mlsanity.com/paper/2601.07006/) (2026)
- [AI as Entertainment](https://mlsanity.com/paper/2601.08768/) (2026)
- [Measuring the Prevalence of Policy Violating Content with ML Assisted Sampling and LLM Labeling](https://mlsanity.com/paper/2602.18518/) (2026)
- [Not All Attention Heads Contribute to Critical Visual Token Selection: Head-Aware Pruning Matters More](https://mlsanity.com/paper/2608.25332/) (2026)
- [Rethinking Ground Truth: A Case Study on Human Label Variation in MLLM Benchmarking](https://mlsanity.com/paper/2603.19744/) (2026)
- [SafeLens: Deliberate and Efficient Video Guardrails with Fast-and-Slow Screening](https://mlsanity.com/paper/2605.17610/) (2026)
- [UNIVID: Unified Vision-Language Model for Video Moderation](https://mlsanity.com/paper/2606.05748/) (2026)
- [Automatic Hard Example Synthesis with Multi-Level Agentic Data Curation](https://mlsanity.com/paper/2607.14256/) (2026)

## Abstract (arXiv)

As the volume of video content online grows exponentially, the demand for moderation of unsafe videos has surpassed human capabilities, posing both operational and mental health challenges. While recent studies demonstrated the merits of Multimodal Large Language Models (MLLMs) in various video understanding tasks, their application to multimodal content moderation, a domain that requires nuanced understanding of both visual and textual cues, remains relatively underexplored. In this work, we benchmark the capabilities of MLLMs in brand safety classification, a critical subset of content moderation for safe-guarding advertising integrity. To this end, we introduce a novel, multimodal and multilingual dataset, meticulously labeled by professional reviewers in a multitude of risk categories. Through a detailed comparative analysis, we demonstrate the effectiveness of MLLMs such as Gemini, GPT, and Llama in multimodal brand safety, and evaluate their accuracy and cost efficiency compared to professional human reviewers. Furthermore, we present an in-depth discussion shedding light on limitations of MLLMs and failure cases. We are releasing our dataset alongside this paper to facilitate future research on effective and responsible brand safety and content moderation.