AI & LLM

Content Moderation Response JSON Example

A JSON example of a content moderation API response — flagged status, per-category booleans, and confidence scores. Copy-ready for OpenAI moderation and trust & safety pipelines.

JSON

Generate mock data Open in Formatter

{
  "id": "modr-9Kd2Lm4Qp",
  "model": "omni-moderation-latest",
  "results": [
    {
      "flagged": false,
      "categories": {
        "harassment": false,
        "hate": false,
        "self-harm": false,
        "sexual": false,
        "violence": false
      },
      "category_scores": {
        "harassment": 0.0004,
        "hate": 0.0001,
        "self-harm": 0.00002,
        "sexual": 0.0003,
        "violence": 0.0007
      }
    }
  ]
}

Field Reference

results[].flaggedrequiredbooleanTrue if any category exceeded its threshold — the headline decision

results[].categoriesrequiredobjectPer-category boolean flags indicating which policies were violated

results[].category_scoresrequiredobjectPer-category confidence scores (0–1); higher means more likely a violation

modelrequiredstringModeration model used to score the content

Variants

Flagged contentA response where the input violated a policy.

Flagged content

Generate mock data Open in Formatter

{
  "results": [
    {
      "flagged": true,
      "categories": {
        "harassment": true,
        "hate": false,
        "violence": false
      },
      "category_scores": {
        "harassment": 0.97,
        "hate": 0.12,
        "violence": 0.05
      }
    }
  ]
}

Common Use Cases

→Screening user-generated content before it's published or sent to an LLM
→Building trust & safety pipelines with auditable decisions
→Setting custom thresholds per category for your risk tolerance

moderationcontent moderationtrust and safetyopenaisafetyai

Validate or format this JSON

One click loads this exact example into the tool — no copy-paste needed. Format it, validate it, explore the tree, or generate TypeScript types instantly.

Format Validate Explore tree Generate TypeScript Minify

Frequently Asked Questions

flagged uses provider default thresholds, which may be stricter or looser than your policy. For production, inspect category_scores and apply your own thresholds per category so you can tune precision vs. recall for your specific audience and risk tolerance.

Moderate both inputs (before sending user text to an LLM, to block prompt abuse) and outputs (before showing model responses to users). Log decisions and scores for auditing and appeals, and avoid storing flagged content longer than necessary.

Related JSON Examples

AI & LLMOpenAI Chat Completion ResponseOpenAI Chat Completions API response with choices, message, finish_reason, and usage.AI & LLMLLM Structured Output (JSON Schema)Structured output request with a JSON Schema and the model's schema-valid response.SocialSocial Media PostSocial post with author info, text content, media, engagement counters, and hashtags.