Transparency Ledger
Every number on this page comes directly from our database. No spin, no hiding, no performative metrics—just honest data about how Dear Nobody is used.
Every Number Reflects a Choice
What to share, what to keep, and what to let go. We honor all three. This page exists because you deserve to know exactly what happens here.
Full Visibility
These numbers update in real-time. What you see here is exactly what's in our database—no delays, no adjustments, no cherry-picking.
Your Control
Every submission can be removed instantly with its unique Private ID. No waiting periods, no approval needed, no questions asked.
No Personal Data
We don't store names, emails, or any identifying information. These stats come from anonymous metadata only.
Submission Statistics
Content Breakdown
Letters
Open letters to anyone
Confessions
Things left unsaid
Explains
Clarifications & context
Moderation Transparency
Public submissions go through human review before appearing in the Mailbox. Private letters are never moderated—they're yours alone.
Publications Over Time
How We Use AI
AI assists our moderators but never makes final decisions. Here's exactly how it works—and how you can verify it.
Inference Only
Our AI analyzes your words. It cannot learn from them. Your submissions are never used to train any AI model.
Human Oversight
AI flags content for attention. Humans make every final decision. No exceptions, no automation of approval.
Your Visibility
You can view exactly what AI detected about your submission using your Private ID. Nothing hidden.
Deletion = Gone
When you delete your submission, all AI analysis is deleted too. Permanently. No backups, no residual data.
AI Accountability Metrics
AI isn't perfect. That — override rate means AI gets it wrong sometimes. We track this so we can improve, and so you know a human always has the final say. We don't show performative metrics like "0 submissions used for training" because that number is always zero by design.
Every Submission's Journey
Here's exactly what happens when you submit a letter.
You Write
Your letter is encrypted in your browser before being sent to our servers.
ECDH-P256 + AES-GCMAI Analyzes
Our AI checks for crisis indicators and suggests themes. This takes about 2 seconds.
Cannot remember after analysisHuman Reviews
A trained moderator reads your letter and makes the final decision.
~— averageDecision Made
Your letter is published, held for clarification, or (rarely) declined.
~— approval rateFor Skeptics
You don't trust companies. Neither do we, honestly. Here's how to verify everything we claim without taking our word for it.
Verify: AI doesn't train on your data
- We use Cloudflare Workers AI — their docs confirm inference-only architecture
- We use Llama — Meta's page shows exactly what it was trained on
- There is no API endpoint for training. It doesn't exist.
Verify: We don't retain your IP
- Open your browser dev tools during submission
- Check the request payload — no IP field
- Our database schema has no IP column
- Infrastructure providers may process network data — see Privacy Policy
Verify: Humans review everything
- Our stats show human override rate
- If AI decided everything, override rate would be 0%
- It's —. That's humans disagreeing with machines.
Still don't trust us?
- Don't use the service (we respect that)
- Request your data via Private ID
- Email us questions — we answer within 48 hours
Activity Over Time
Monthly Submissions & Removals
Visibility Choice
How We Calculate These Numbers
We compute these statistics from non-personally-identifiable metadata stored in our database. Here's what we track:
No personal data is processed or displayed. We don't track who writes what—only aggregate counts. Private letters are never reviewed or read by moderators.