Learning objectives
By the end of this lesson, you will be able to:
- Identify the AI-powered features added to the Comments screen when Comment Moderation and Suggest Reply are enabled
- Explain the automatic moderation threshold Comment Moderation applies to incoming comments
- Apply bulk sentiment and toxicity analysis to an existing comment queue and interpret the results
- Evaluate when to adjust the default moderation threshold and how tone selection affects Suggest Reply output
Enabling comment AI features
Comment Moderation and Suggest Reply are both in Settings > AI > Admin Experiments. Enable each toggle individually.
Before enabling, confirm that your site has a connected AI provider in Settings > Connectors. Both features require a text-generation-capable model to function.
💡 Why this matters for you: If Connector Approval is also enabled on your site, AI features will not work until the AI Plugin has been approved in Tools > Connector Approvals. If Comment Moderation does not appear to be running after you enable it, check your Connector Approval status first. Connector Approval is covered in [Lesson 3: Monitoring and controlling AI on your site](Link — verify URL before publishing).
📸 Screenshot placeholder: Settings > AI page scrolled to the Admin Experiments section, showing Comment Moderation and Suggest Reply toggles enabled.
What Comment Moderation does
When Comment Moderation is enabled, two things happen:
New comments are analyzed automatically. When a comment is submitted, the plugin analyzes it for toxicity and sentiment. If a comment meets both thresholds — a toxicity score of 0.7 or higher and a negative sentiment rating — it is automatically moved to pending moderation for your review. Comments that do not meet both conditions are not automatically held.
The Comments screen gains two new columns. Navigate to Comments > All Comments. You will see a Sentiment column and a Toxicity column alongside each comment. Each comment shows a badge reflecting its current analysis state: pending analysis, analyzed with results, or failed.
To analyze comments that were submitted before you enabled Comment Moderation, use the bulk action. Select the comments you want to analyze, choose Analyze Sentiment and Toxicity from the Bulk actions menu, and click Apply. The plugin queues the selected comments and processes them sequentially in your browser. You will see the badges update as each comment is analyzed.
📸 Screenshot placeholder: Comments screen showing the Sentiment and Toxicity columns, with badges visible on analyzed comments and the Analyze Sentiment and Toxicity bulk action selected in the Bulk actions menu.
Reading the analysis results
Toxicity scores run from 0 to 1. A score of 0.7 or higher is the default threshold for automatic moderation action. Sentiment is classified as positive, negative, or neutral.
The automatic rule requires both conditions to be true: a comment must have high toxicity AND negative sentiment to be moved to pending automatically. A comment that is negative in sentiment but low in toxicity — a disappointed but civil response, for example — will not be automatically held.
When you review comments in your queue, use the badges as a starting signal, not a final verdict. Read the comment in full before deciding to approve, mark as spam, or trash it.
💡 Why this matters for you: The 0.7 toxicity threshold is a default, not a requirement. If your site’s community standards call for a lower threshold, a developer can adjust this using the plugin’s filter hooks. Flag this for your developer if you need a stricter or more lenient rule.
Suggesting replies
When Suggest Reply is enabled, a Suggest reply action appears when you hover over any comment in the Comments screen. It also appears in the Activity widget on the WordPress Dashboard.
Clicking Suggest reply automatically opens the inline reply editor and inserts an AI-generated reply into the text field. You do not need to open the editor separately. From there, you have two options before posting:
Edit the reply directly. The suggestion is pre-filled text — treat it as a starting draft and adjust it to match your voice.
Change the tone and regenerate. A tone dropdown in the inline editor offers three options: friendly (the default), professional, and casual. Select a different tone and click the regenerate button to get a new suggestion with that tone applied.
When you are happy with the reply, post it as you normally would.
📸 Screenshot placeholder: Comments screen showing the inline reply editor open with an AI-generated reply pre-filled and the tone dropdown visible with friendly, professional, and casual options.
Building a practical moderation workflow
A sustainable comment moderation workflow with these tools might look like this:
For new comments: Let automatic moderation catch high-risk comments and move them to pending. Review the pending queue regularly, reading each held comment before deciding whether to approve or discard.
For existing backlogs: Use the bulk analysis action to analyze comments submitted before you enabled Comment Moderation. Work through flagged results in batches rather than trying to process everything at once.
For replies: Use Suggest Reply as a drafting tool, not a publishing shortcut. Choose the tone that fits the conversation, edit for your voice, and post with intention.
Key terms
| Term | Definition |
| Comment Moderation | An Admin Experiment that adds sentiment and toxicity analysis to the Comments screen and automatically holds comments that exceed the moderation threshold. |
| Toxicity score | A number between 0 and 1 indicating how much hostile or harmful language a comment contains. The default automatic moderation threshold is 0.7. |
| Sentiment | The overall tone of a comment: positive, negative, or neutral. Used alongside the toxicity score to determine automatic moderation. |
| Bulk action | A WordPress feature that applies an action to multiple selected items at once. Comment Moderation adds the Analyze Sentiment and Toxicity bulk action to the Comments screen. |
| Suggest Reply | An Admin Experiment that adds a Suggest reply action to the Comments screen and Activity widget, generating a pre-filled draft reply with tone controls. |
| Tone | In Suggest Reply, the style applied to generated replies: friendly (default), professional, or casual. |
✅ Check your understanding
- A comment arrives with a toxicity score of 0.65 and a negative sentiment rating. It does not get held automatically. You read it and find it contains genuinely hostile language. What does this tell you about the automatic moderation threshold, and what would you do in this situation?
- You use Suggest Reply on a comment from a long-time reader. The default friendly tone feels right, but the generated reply is longer than you want. What is your next step before posting?