Most WordPress comment moderation is cleanup after the mess lands.
Someone posts a threat, an insult, identity-based hate, or sexually explicit language. It hits wp-comments-post.php, lands in your database, and sits in the moderation queue until you (or a plugin) notice. By then you have already stored abuse you never wanted, and a human has to decide what to do with it.
That workflow assumes the comment should reach you first. Generatify Content Moderation flips the order: it checks the comment as it is being written, and blocks submission when the model flags toxic content. The commenter is told the comment cannot be submitted and can edit it. Nothing reaches your moderation queue. Nothing reaches your database.
This post stays on that feature. Generatify also ships a helpdesk agent and a page agent for other jobs (we covered those in how to set up an AI helpdesk and why a chatbot bubble is not enough). Today is about stopping toxic comments at the form, not answering support tickets.
The problem with “approve later”
WordPress Discussion settings are solid for after-the-fact control. Hold comments for review. Hold comments that contain certain words. Hold comments from first-time commenters. Those rules still matter, and you should keep them.
What they do not do is stop the submit. Abusive text still gets written into your database, still shows up in your queue, and still costs attention. On a busy blog, that queue fills with heat-of-the-moment insults you would rather never store. On a quieter site, one ugly comment still sits there until you open the dashboard.
Server-side AI moderation tools usually send the comment (or a hash of it) to a provider, burn API credits, and still run after submit. That is a different product shape. Generatify’s toxicity check is client-side on purpose, which changes cost, privacy, and what you can honestly claim about it.
What Generatify Content Moderation actually checks
Path in the plugin: Generatify > Settings > Content Moderation.
When AI comment moderation is on, Generatify checks a comment as it is being written and blocks submission if it detects toxic content in these categories:
- Threats
- Insults
- Obscenity
- Identity-based hate
- Sexually explicit language
That list comes from the model the feature was trained against. It is not a custom keyword list you maintain by hand. It is also not a general “is this comment good for my brand?” judge.
If the check fires, the commenter sees that the comment cannot be submitted and can edit the text. From your side, the important part is what does not happen: no row in the comments table, no pending item in the queue, no email about a new comment that is pure abuse.
Why the model runs in the commenter’s browser
The docs are blunt about three consequences of running the model on the visitor’s device.
No cost, no provider. There is no API key for this feature and no per-comment charge, however many comments you get. That is separate from Generatify Connectors. ChatGPT, Claude, Gemini, and Kimi keys power other AI features (editor generation, helpdesk conversation mode, and similar). Content Moderation does not need those keys.
No content leaves the visitor’s device. The comment is scored locally. It is not sent to a third party just to decide whether it is toxic. For sites that care about comment privacy, that is the headline, not a footnote.
A first-load download, then cache. The model is fetched the first time a commenter reaches a page with comments. After that it is cached. First visit to a comment form may feel a touch heavier while the model lands. Later visits reuse what the browser already has.
If you have been avoiding “AI moderation” because every quote came with a usage meter, this is the feature that breaks that pattern.
Usability feature, not a security control
Read this twice before you turn it on and declare the war on spam over.
Anything that posts comments without running your JavaScript is unaffected. A bot that posts straight to wp-comments-post.php never runs the check. Generatify’s docs call this a usability feature, not a security control. It stops a real person writing something abusive in the heat of the moment. It does not stop automated spam.
Keep your existing spam protection. Akismet, Antispam Bee, Cloudflare rules, honeypots, Discussion keyword holds: whatever already works on your site stays on. Toxicity moderation reduces abusive human submits. It does not replace anti-spam.
What it does not catch
Be honest with yourself about the gaps, or you will blame the plugin for jobs it never claimed.
It does not detect:
- Spam
- Off-topic comments
- Link dropping
- Politely worded bad faith
Detection is mainly in English. Toxic comments in other languages will mostly pass through. If your community is multilingual, treat this as an English-layer filter, not a global one.
Expect occasional false positives. Quoted abuse, reclaimed slurs, and frank discussion of a sensitive topic can all look toxic to a classifier. Because the block happens before submission, a false positive means a legitimate commenter simply cannot post, and they usually will not email you about it. On a site with heated but legitimate discussion (politics, trauma support, hard news), weigh that carefully against what you gain. On a product blog, tutorial site, or store with light comments, the trade is usually easier.
Pair it with WordPress Discussion settings
The toxicity checker and WordPress moderation solve different halves of the same problem. The docs put it in a table that is worth internalizing:
| | Toxicity checker | WordPress moderation | | — | — | — | | When | Before submission | After submission | | Stops | Abusive language | Anything matching your rules | | Visible to the commenter | Yes, immediately | No | | Works without JavaScript | No | Yes |
Run both. Under Settings > Discussion, keep your keyword list and your held-for-review rules. The toxicity checker reduces what reaches those rules. It does not replace them.
A practical stack that matches the docs:
- Turn on Generatify Content Moderation for English toxic language at the form.
- Keep Discussion keyword holds for brand-specific terms, competitor spam phrases, and anything your community flags that is not “toxicity” in the model sense.
- Keep a real spam plugin for bots that skip JavaScript.
- Still review held comments. Human judgment stays in the loop for edge cases.
That combination is boring. Boring is what you want for comment ops.
How this fits the rest of Generatify (without drifting the topic)
Generatify is an AI toolkit for WordPress: automations, agents, smart suggestions, connectors, and moderation. Content Moderation is the quiet one. No provider key. No queue of AI drafts. Just a browser-side gate on the comment form.
If you already use the helpdesk for support questions or the page agent for navigation on complex sites, moderation is still worth enabling on its own. Those features serve visitors who need answers or actions. This one protects the people who leave comments and the people who moderate them.
Connectors still matter for the rest of the product. When you want editor-side generation or conversation-mode helpdesk, you connect ChatGPT, Claude, Gemini, or Kimi with your own key. That setup is unrelated to comment toxicity. You can ship moderation without wiring a provider at all.
Membership note (because Pro is how you unlock the full stack)
DaftPlug sells Pro as one membership, not five separate licenses. One Pro plan unlocks Generatify along with Progressify, Lightify, Rankify, and Snapshotify. Plans differ by site count, not by which plugins you get. New Pro plugins join existing memberships without a new line item.
Compare plans on daftplug.com/pricing. Free Generatify on WordPress.org is fine for trying the shape of the product. Pro is what you want when you are running the full AI surface, including the pieces that need a license.
Turn it on without overselling yourself
A sane rollout looks like this:
- Open Generatify > Settings > Content Moderation and enable AI comment moderation.
- Leave spam plugins and Discussion holds alone. Do not delete keyword lists because “AI will handle it.”
- Post a test comment with clear insult language on a staging page (or a private post) and confirm the form blocks submit and lets you edit.
- Post a normal, polite comment and confirm it still goes through.
- If you run heated legitimate threads, watch for false positives in the first week and decide whether the trade still makes sense for those sections.
If a commenter reports they “cannot post,” ask what they tried to say before you assume the site is broken. Quoted abuse and sensitive frank talk are the usual false-positive cases from the docs.
Bottom line
Toxic WordPress comments do not have to land in your queue first. Generatify Content Moderation checks the text in the commenter’s browser for threats, insults, obscenity, identity-based hate, and sexually explicit language, then blocks submit before anything hits your database. No provider key. No per-comment fee. No third-party scoring hop.
It will not catch spam bots, link droppers, or polite bad faith. Keep those defenses. Pair the checker with Settings > Discussion. Be careful on communities where frank talk is the point.
If you want that gate on your comment forms, start with Generatify, keep your spam stack, and unlock the rest of the DaftPlug Pro membership from pricing when you are ready to run AI across the site without buying five separate tools.