# Can Foundation Models Moderate Online Content? Evaluating Instruction- vs. Example-Driven Policy Operationalization

arXiv: https://arxiv.org/abs/2609.10410 · HTML: https://mlsanity.com/paper/2609.10410/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 18 of its 49 references that are in the map.
- First posted on arXiv 9 Sep 2026.

## Builds on (most cited first, 18 of 18)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [LLaVA-OneVision: Easy Visual Task Transfer](https://mlsanity.com/paper/2408.03326/) (2024)
- [Billion-scale similarity search with GPUs](https://mlsanity.com/paper/1702.08734/) (2017)
- [InternVL3.5: Advancing Open-Source Multimodal Models in Versatility, Reasoning, and Efficiency](https://mlsanity.com/paper/2508.18265/) (2025)
- [The Hateful Memes Challenge: Detecting Hate Speech in Multimodal Memes](https://mlsanity.com/paper/2005.04790/) (2020)
- [HateXplain: A Benchmark Dataset for Explainable Hate Speech Detection](https://mlsanity.com/paper/2012.10289/) (2020)
- [Social Bias Frames: Reasoning about Social and Power Implications of Language](https://mlsanity.com/paper/1911.03891/) (2019)
- [Qwen3-VL-Embedding and Qwen3-VL-Reranker: A Unified Framework for State-of-the-Art Multimodal Retrieval and Ranking](https://mlsanity.com/paper/2601.04720/) (2026)
- [Evaluating the World Model Implicit in a Generative Model](https://mlsanity.com/paper/2406.03689/) (2024)
- [HateMM: A Multi-Modal Dataset for Hate Video Classification](https://mlsanity.com/paper/2305.03915/) (2023)
- [Policy-as-Prompt: Rethinking Content Moderation in the Age of Large Language Models](https://mlsanity.com/paper/2502.18695/) (2025)
- [MLLM-as-a-Judge for Image Safety without Human Labeling](https://mlsanity.com/paper/2501.00192/) (2024)
- [The DSA Transparency Database: Auditing Self-reported Moderation Actions by Social Media](https://mlsanity.com/paper/2312.10269/) (2023)
- [Bluesky and the AT Protocol: Usable Decentralized Social Media](https://mlsanity.com/paper/2402.03239/) (2024)
- [Automated Transparency: A Legal and Empirical Analysis of the Digital Services Act Transparency Database](https://mlsanity.com/paper/2404.02894/) (2024)
- [A Year of the DSA Transparency Database: What it (Does Not) Reveal About Platform Moderation During the 2024 European Parliament Election](https://mlsanity.com/paper/2504.06976/) (2025)
- [The Gray Area: Characterizing Moderator Disagreement on Reddit](https://mlsanity.com/paper/2601.01620/) (2026)
- [Prompt Governance? On Governing Technologies Governed by Natural Language](https://mlsanity.com/paper/2606.07539/) (2026)

## Built on by (most cited first, 1 of 1)

- [Characterizing Bluesky Content Moderation Service: From Automation of Service to Landscape of Harms](https://mlsanity.com/paper/2609.11373/) (2026)

## Abstract (arXiv)

The growing complexity of content moderation policies presents a critical challenge for their consistent operationalization. While foundation models possess the basic capabilities needed to confront this challenge, whether they can reliably moderate online content remains an unanswered question. In this paper, we systematically compare two competing paradigms for Vision-Language Model (VLM) guidance: an instruction-driven approach where models reason from policy precepts, and an example-driven approach where they generalize from prior precedents. We ground this investigation in ModerationBench, a new benchmark of 4,000 manually annotated, in-the-wild posts from the Bluesky platform. Our experiments reveal that foundation models can substantially outperform Bluesky's deployed moderation system, nearly tripling its F1 score (0.60 vs. 0.22) on Random Posts in the benchmark, with both instruction- and example-driven paradigms achieving comparable peak effectiveness. Our findings thus chart a path toward reliable and adaptable policy operationalization at scale.