mlsanityData to 7 Oct 2026

BLM-Guard: Explainable Multimodal Ad Moderation with Chain-of-Thought and Policy-Aligned Rewards

arXiv 2602.18193 ↗First posted 20 Feb 2026cs.CVcited by 2

On the map: LLM reasoning → Multimodal reasoning

Abstract (arXiv)

Short-video platforms now host vast multimodal ads whose deceptive visuals, speech and subtitles demand finer-grained, policy-driven moderation than community safety filters. We present BLM-Guard, a content-audit framework for commercial ads that fuses Chain-of-Thought reasoning with rule-based policy principles and a critic-guided reward. A rule-driven ICoT data-synthesis pipeline jump-starts training by generating structured scene descriptions, reasoning chains and labels, cutting annotation costs. Reinforcement learning then refines the model using a composite reward balancing causal coherence with policy adherence. A multitask architecture models intra-modal manipulations (e.g., exaggerated imagery) and cross-modal mismatches (e.g., subtitle-speech drift), boosting robustness. Experiments on real short-video ads show BLM-Guard surpasses strong baselines in accuracy, consistency and generalization.

Key facts. Cited by 2 papers in the map (data through 7 Oct 2026). Placed in Multimodal reasoning, one of 17 directions in LLM reasoning. Builds on 24 of its 32 references that are in the map. First posted on arXiv 20 Feb 2026.

Builds on

24 of 32 references.

Closest ancestors · 3

Same direction, newest first

  1. 2025GRPO-CARE: Consistency-Aware Reinforcement Learning for Multimodal Reasoning19 Jun
  2. 2025Kimi-VL Technical Report10 Apr
  3. 2024Interleaved-Modal Chain-of-Thought29 Nov

Other references · 21

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,456
  2. 2025Qwen2.5-VL Technical Reportcited by 5,418
  3. 2023Tree of Thoughts: Deliberate Problem Solving with Large Language Modelscited by 3,155
  4. 2022Self-Instruct: Aligning Language Models with Self-Generated Instructionscited by 2,674
  5. 2025InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Modelscited by 1,769
  6. 2023Llama Guard: LLM-based Input-Output Safeguard for Human-AI Conversationscited by 1,124
  7. 2023Generative Language Models and Automated Influence Operations: Emerging Threats and Potential Mitigationscited by 157
  8. 2025Adaptive Keyframe Sampling for Long Video Understandingcited by 153
  9. 2024GuardAgent: Safeguard LLM Agents by a Guard Agent via Knowledge-Enabled Reasoningcited by 113
  10. 2025GuardReasoner: Towards Reasoning-based LLM Safeguardscited by 89
  11. 2025MM-RLHF: The Next Step Forward in Multimodal LLM Alignmentcited by 79
  12. 2024UnsafeBench: Benchmarking Image Safety Classifiers on Real-World and AI-Generated Imagescited by 52
9 more
  1. 2024LlavaGuard: An Open VLM-based Framework for Safeguarding Vision Datasets and Modelscited by 49
  2. 2024On the Hardness of Faithful Chain-of-Thought Reasoning in Large Language Modelscited by 42
  3. 2022FakeSV: A Multimodal Benchmark with Rich Social Context for Fake News Detection on Short Video Platformscited by 41
  4. 2025GuardReasoner-VL: Safeguarding VLMs via Reinforced Reasoningcited by 38
  5. 2024SafeWatch: An Efficient Safety-Policy Following Video Guardrail Model with Transparent Explanationscited by 37
  6. 2025VLM as Policy: Common-Law Content Moderation Framework for Short Video Platformcited by 19
  7. 2024Moderating Illicit Online Image Promotion for Unsafe User-Generated Content Games Using Large Vision-Language Modelscited by 13
  8. 2024T2Vs Meet VLMs: A Scalable Multimodal Dataset for Visual Harmfulness Recognitioncited by 10
  9. 2025MSR-Align: Policy-Grounded Multimodal Alignment for Safety-Aware Reasoning in Vision-Language Modelscited by 7

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 2.

2 papers.

  1. 2026RuleSafe-VL: Evaluating Rule-Conditioned Decision Reasoning in Vision-Language Content Moderationcited by 1
  2. 2026ARGUS: Policy-Adaptive Ad Governance via Evolving Reinforcement with Adversarial Umpiringcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Summarize, Judge, Refine: Decoupled Content Understanding and Policy Learning for Multimodal Content Moderation
  2. 2025RAVEN++: Pinpointing Fine-Grained Violations in Advertisement Videos with Active Reinforcement Reasoning
  3. 2025RAVEN: Robust Advertisement Video Violation Temporal Grounding via Reinforcement Reasoning
  4. 2026Beyond Single Videos: Benchmarking and Active Evidence Seeking for E-Commerce Cross-Video Reasoning
  5. 2025AI vs. Human Moderators: A Comparative Evaluation of Multimodal LLMs in Content Moderation for Brand Safety
  6. 2026Adapting Reinforcement Learning with Chain-of-Thought Supervision for Explainable Detection of Hateful and Propagandistic Memes
  7. 2026FactGuard: Agentic Video Misinformation Detection via Reinforcement Learning
  8. 2025Towards Trustworthy Multimodal Moderation via Policy-Aligned Reasoning and Hierarchical Labeling