Skip to main content

AI Content Moderation

Better Messages offers AI-powered content moderation to automatically detect and handle harmful content. Choose between two moderation providers depending on your needs.

Moderation Providers#

Better Messages Moderation AI#

Included with the WebSocket license at no additional cost. No third-party API keys required.

  • Powered by Better Messages Cloud
  • 23 content categories including extended categories not available in OpenAI
  • Website context — describe what your site is used for so moderation adapts to your community norms
  • Custom moderation rules — define your own rules in plain text
  • Conversation context awareness — detect patterns split across multiple messages (e.g. phone numbers sent in separate messages)
  • Text and image moderation

OpenAI Moderation API#

Free to use with an OpenAI API key.

  • Requires an OpenAI API key configured in Integrations → OpenAI
  • Fixed set of content categories
  • Configurable confidence threshold
  • Text and image moderation

How it works#

When enabled, messages are analyzed by the selected moderation provider. You can choose between two actions when harmful content is detected:

  • Flag Only (Recommended) — Message is delivered normally but marked for admin review. Best for user experience since no AI is 100% accurate and legitimate messages won't be blocked.
  • Hold for Review — Message is held until an admin approves or rejects it. More strict, but may delay legitimate messages.

Moderators can review flagged and held messages in the admin Messages Viewer.

Content Categories#

Both providers support the following base categories:

  • Hate / Hate Threatening
  • Harassment / Harassment Threatening
  • Sexual Content / Sexual Minors
  • Violence / Violence Graphic
  • Self-Harm / Self-Harm Intent / Self-Harm Instructions
  • Illicit / Illicit Violent

Better Messages Moderation AI adds these extended categories:

  • Spam
  • Scam / Phishing
  • Minor Safety
  • Contact Sharing
  • Profanity
  • Impersonation
  • Doxxing
  • Drugs / Alcohol
  • Threats
  • Commercial Promotion

Selecting a parent category automatically covers its subcategories.

Better Messages AI — Additional Features#

Website Context#

An optional free-text description of what your website is used for. Without it, the AI judges messages by generic defaults, which can feel too aggressive on sites where the "borderline" content is exactly what members are there for. With it, the AI calibrates moderation to what is normal and expected between your members.

Website context is especially useful for:

  • Dating and matchmaking sites — flirting, compliments and romantic conversation are normal, not harassment
  • Gaming and esports communities — competitive trash talk is not a threat
  • Adult communities — consensual conversation between adult members is expected
  • Support and recovery communities — frank discussion of sensitive topics is the community's purpose, not a violation

Example:

This is a dating site for adults. Flirting, compliments and romantic conversation between members are normal and expected here.

If your site is a general-purpose community, you can leave this field empty — the default moderation behavior already fits.

The website context calibrates strictness only — your selected content categories and custom rules still apply, and content involving minors is never excused.

Custom Rules#

Define your own moderation rules in plain text, one rule per line. For example:

Block contact info (phones, emails, social handles) and off-platform moves
Block promotions, affiliate links, recruitment

Custom rules and content categories work independently. You can use either one alone or combine both.

Using only custom rules#

The predefined content categories are intentionally general. If they don't match your moderation policy, uncheck all content categories and define only custom rules. The AI then works as a strict pattern matcher: a message is flagged only when one of your rules is violated — tone, profanity and anything else not covered by a rule is ignored. This is the most precise way to match your exact preferences.

You can combine this with Website Context so your rules are interpreted with your site in mind — for example, a rule like "block attempts to move the conversation off-platform" is judged differently on a dating site than on a support forum.

Conversation Context#

Set the number of previous messages to include for context (0–20). This helps detect patterns like phone numbers or emails split across multiple messages. Recommended: 5–10.

How to enable#

Navigate to WP Admin → Better Messages → Settings → Moderation.

  1. Select a Moderation Provider
  2. Enable AI Moderation
  3. Choose the Flagged Message Action
  4. Select Content Categories to detect — or uncheck them all to rely on custom rules only
  5. Optionally enable Moderate Images
  6. Set Bypass Roles for trusted user roles
Data Privacy

Message content is sent to the selected provider for analysis. Better Messages Cloud does not store any message data — content is analyzed in real time and immediately discarded. If using OpenAI, please review OpenAI's data usage policy.

Requirements
  • Better Messages Moderation AI requires a WebSocket license
  • OpenAI Moderation API requires PHP 8.1+ and an OpenAI API key