Important caveat


TL;DR — the privacy posture in one paragraph


1. What data the plugin processes

Sent to your AI provider on every reply generation

Data Source Notes
Comment text The visitor's submission Run through wp_strip_all_tags and html_entity_decode first
Post title The post being commented on
Post excerpt First 40 words of post content HTML stripped
Post author display name wp_users.display_name
Post author bio First 30 words of description user meta
Tone, length, language instructions Generated from your settings Static text, not user data

Stored in your WordPress database (never leaves your site)

Data Storage location
API key Option replymind_api_key
Plugin settings Option replymind_settings
Generation log entries Option replymind_logs (last 200, autoload off)
Per-comment AI status flags Comment meta keys prefixed _replymind_*
AI-generated reply text (drafts) Comment meta _replymind_pending_reply

Sent to ReplyMind's servers

Nothing. ReplyMind has no servers in the loop. The plugin is a direct client of OpenAI's and Anthropic's APIs.

Not collected at all

  • Visitor IP addresses are not sent to the AI.
  • Visitor email addresses are not sent to the AI.
  • Visitor names are not sent to the AI.
  • User-agent / browser fingerprint is not sent to the AI.
  • Cookies set by your site are not sent to the AI.

The AI sees the visitor's words, the post they commented on, and your tone preferences — nothing else.


2. WordPress privacy framework integration

ReplyMind registers itself with WordPress's built-in privacy framework on admin_init:

Privacy Policy Guide entry

A notice is added to Tools → Privacy → Policy Guide automatically when the plugin is active. Site administrators can copy this language into their privacy policy:

Personal data exporter

When a user invokes their data-export right (Tools → Export Personal Data with their email), ReplyMind:

  • Looks up all comments authored by that email address.
  • For each comment with AI metadata, exports:
  • Pending AI reply text (if a draft exists)
  • AI reply published marker (if a reply was already posted)
  • Paginates 100 comments per page, returns proper done flags.

Personal data eraser

When a user invokes their right to be forgotten:

  • ReplyMind looks up all their comments.
  • Removes the meta keys _replymind_pending_reply, _replymind_error, _replymind_processed, and _replymind_reply_generated for each.
  • Paginates 100 comments per page.

The original comments themselves are not deleted — that's WordPress's job, not the plugin's.


3. Skipping specific comments programmatically

The replymind_send_comment_data filter lets you exclude individual comments from being sent to the AI at all. Useful for:

  • Sensitive topics or specific posts.
  • Particular author email domains.
  • Comments containing certain keywords.
// Skip all comments on posts in a specific category
add_filter( 'replymind_send_comment_data', function ( $allow, $comment ) {
    if ( has_category( 'sensitive', $comment->comment_post_ID ) ) {
        return false;
    }
    return $allow;
}, 10, 2 );
// Skip comments from a specific email or domain
add_filter( 'replymind_send_comment_data', function ( $allow, $comment ) {
    $excluded_domains = array( '@internal.example.com', '@private.org' );
    foreach ( $excluded_domains as $domain ) {
        if ( str_ends_with( $comment->comment_author_email, $domain ) ) {
            return false;
        }
    }
    return $allow;
}, 10, 2 );

When this filter returns false, the comment is never sent to the AI provider — no draft is generated, no API call is made, no log entry is recorded. The comment is treated as if the plugin doesn't exist.

The filter is checked in three code paths: 1. The cron handler (new comments) 2. The per-comment Generate AI Reply admin action 3. The batch process loop

All three respect the same return value.


4. Uninstall behaviour

When an administrator deletes the plugin from Plugins → Installed Plugins → Delete, uninstall.php runs and removes:

  • Option replymind_api_key
  • Option replymind_settings
  • Option replymind_logs
  • Option replymind_do_activation_redirect
  • All comment meta keys starting _replymind_ (uses LIKE '_replymind_%' with proper prepared-statement escaping)

The AI reply comments themselves are not deleted — they're regular WordPress comments and the plugin doesn't track which it authored beyond the _replymind_reply_generated meta flag (which is itself deleted on uninstall).

If you also want to delete the AI reply comments, you'll need to do that manually before uninstalling, or through a separate cleanup script.


5. Data Processing Agreements (DPAs)

Under GDPR Article 28, if a third party processes personal data on your behalf, you typically need a Data Processing Agreement with them.

With ReplyMind: No DPA needed with the plugin author. We don't process or receive any visitor data.

With your AI provider: A DPA may be required, depending on the personal data your comments contain and your jurisdiction. Both major providers offer DPAs:

Whether your specific use needs a signed DPA depends on the scope of personal data in your comments and your local interpretation of GDPR. Ask your DPO or lawyer.


6. Data residency

You select your AI provider; their data residency rules apply.

  • OpenAI: Primary processing in the United States. EU data residency is offered for some Enterprise plans — see OpenAI's docs.
  • Anthropic: Primary processing in the United States. AWS Bedrock and Google Cloud Vertex AI integrations let you run Claude in EU regions if you switch providers (note: ReplyMind's free plugin uses Anthropic's direct API, not Bedrock).

If your compliance posture requires EU-only processing of comment data, you'll need to choose a provider and tier that supports that — the plugin sends data wherever the provider's API hosts it.


7. What you (the site owner) need to do

Even with everything above, the site owner is the data controller and remains responsible for:

a) Disclosure in your privacy policy

Tell visitors that you use AI to assist with comment replies and that comment text may be sent to a third-party processor. Example wording:

b) Lawful basis

Identify your lawful basis for processing (typically legitimate interest for transactional comment moderation; sometimes consent if you're operating in stricter jurisdictions). Document it.

c) Provider compliance

Verify that your selected AI provider's terms, security posture, and data handling meet your compliance requirements.

d) Block what shouldn't be sent

Use the replymind_send_comment_data filter (above) for any category of comment that shouldn't be processed by an AI service — e.g. comments on posts about medical conditions, financial details, or other sensitive topics.


8. CCPA, LGPD, PIPEDA, and other regimes

The plugin's design — minimal data sent, no plugin-author servers, full export/erase — supports compliance with most major privacy regimes. Specific obligations vary; check with a lawyer in your operating jurisdiction.


9. Where to read more


Read next