Babel Shield
Babel Shield is an AI-powered content moderation API. Send any text content to the API, receive category-level scores in real time, and let your application decide what action to take.
What Problems Does It Solve?
User-generated content is unpredictable. Comment sections, chat messages, form submissions, and reviews can contain spam, hate speech, profanity, scams, and other harmful material. Reviewing this content manually doesn't scale.
Babel Shield analyzes text content and returns a confidence score (0-100) for each moderation category. Your application uses these scores to automatically approve, flag for review, or block content based on thresholds you control.
Key Features
- 13 moderation categories covering spam, profanity, hate speech, harassment, violence, sexual content, self-harm, deception, low-quality content, crypto scams, illegal content, prompt injection, and generally offensive material
- Real-time scoring on a 0-100 scale for each category
- Configurable thresholds so you decide what gets flagged -- strict for children's platforms, relaxed for adult communities
- Response caching for identical content, reducing costs and speeding up responses
- Multi-user accounts with role-based access control -- owners, operators, and viewers each see what they need
- Host-restricted tokens for added security when calling the API from browser-based applications
How It Works
- Send content to the API with a single HTTP request
- Receive category scores back instantly -- a confidence score (0-100) for each of the 13 moderation categories
- Check the
flaggedfield to see if any category exceeded its threshold, then decide what your application should do
Flagging is advisory, not enforcement. Your application decides whether to approve, queue for human review, or block flagged content.
Moderation Categories
| Category | API Key | Description | Example Triggers |
|---|---|---|---|
| Spam & Marketing | spam |
Unwanted promotional or repetitive content | SEO spam, phishing links, "buy now" pressure, bulk messages |
| Profanity | profanity |
Vulgar language and explicit words | Swearing, slurs, crude language |
| Sexual Content | sexual |
Adult or sexually explicit content | Explicit descriptions, suggestive material |
| Offensive | offensive |
Generally offensive or inappropriate content | Insults, slurs, demeaning language |
| Violence | violence |
Content depicting or promoting violence | Threats, graphic descriptions, incitement |
| Hate Speech | hate |
Content promoting hatred against groups | Discrimination, supremacist language, dehumanization |
| Harassment | harassment |
Targeted harassment or bullying content | Personal attacks, intimidation, doxxing |
| Self-Harm | self-harm |
Content related to self-harm or suicide | Suicide methods, self-injury encouragement |
| Deception | deception |
Misleading or fraudulent content | Misinformation, impersonation, fake claims |
| Low Quality | garbage |
Low-quality or nonsensical content | Gibberish, random characters, fake form data |
| Crypto & Financial Fraud | crypto |
Cryptocurrency scams and financial schemes | Pump-and-dump schemes, fake ICOs, wallet scams |
| Illegal Content | illicit |
Content promoting illegal activities | Drug sales, weapons trafficking, dangerous goods |
| Prompt Injection | injection |
Attempts to override or manipulate the AI moderation instructions | "Ignore previous instructions", jailbreak prompts |
Next Steps
- Quick Start -- Add Babel Shield to your website with one script tag
- Examples & Guides -- Choose the right guide for your platform and experience level
- Getting Started -- Set up your account and moderate your first content in minutes
- Configuration Reference -- All configuration options, thresholds, feedback modes, and proxy setup
- API Reference -- Complete endpoint documentation with code examples