Open-Domain Safety Policy Construction

TL;DR

This paper introduces Deep Policy Research (DPR), a system that automatically drafts domain-specific content moderation policies using web search and minimal human input, outperforming baselines and matching expert quality.

Contribution

DPR is a novel, minimal agentic system that efficiently generates comprehensive moderation policies from seed information, reducing costs and outperforming general research methods.

Findings

DPR outperforms definition-only and in-context learning baselines.

DPR is competitive with expert-written policies in several domains.

DPR surpasses a general-purpose deep research system under the same conditions.

Abstract

Moderation layers are increasingly a core component of many products built on user- or model-generated content. However, drafting and maintaining domain-specific safety policies remains costly. We present Deep Policy Research (DPR), a minimal agentic system that drafts a full content moderation policy based on only human-written seed domain information. DPR uses a single web search tool and lightweight scaffolding to iteratively propose search queries, distill diverse web sources into policy rules, and organize rules into an indexed document. We evaluate DPR on (1) the OpenAI undesired content benchmark across five domains with two compact reader LLMs and (2) an in-house multimodal advertisement moderation benchmark. DPR consistently outperforms definition-only and in-context learning baselines, and in our end-to-end setting it is competitive with expert-written policy sections in several domains. Moreover, under the same seed specification and evaluation protocol, DPR outperforms a general-purpose deep research system, suggesting that a task-specific, structured research loop can be more effective than generic web research for policy drafting. We release our experiment code at https://github.com/xiaowu0162/deep-policy-research.