Privacy

Your original conversation is never saved

When you upload screenshots or paste text, it's sent securely to our AI provider (Anthropic) to generate your verdict, and used only in-memory for that single request. Your pasted conversation text is not saved to our database in any form — not in full, and not as a preview or excerpt. Screenshots are never stored either — they're used only for that one analysis and then discarded. We save only the structured, non-sensitive result of the analysis (see "What we store" below) so you can revisit the case in your history. Your public share card and public share page never show your original conversation or screenshots — they only ever display participant names (or "Person A" / "Person B"), the responsibility percentages, the verdict categories, the AI Judge one-liner, and Who's Wrong? branding.

What we store

• Your account email and (hashed, never plaintext) password
• The relationship type, participant names, and structured verdict for each case
• A non-sensitive description of the input shape only (e.g. "Text conversation" or "3 screenshots") — never any portion of the actual conversation content
• The generated verdict and, if you use it, Roast Mode text
• Case timestamps
• Usage counts, to enforce the free daily limit
• Billing information, handled entirely by Stripe — we never see your card number

Deleting your data

You can permanently delete any saved case at any time from your history page. Deletion is immediate and irreversible.

Anonymous use

You can try Who's Wrong? without creating an account. We use a private, non-identifying cookie solely to enforce the 3-cases-per-day free limit. No case content is tied to your identity unless you create an account.

Third parties

We use Anthropic's API to analyze cases, and Stripe to process payments. Neither is permitted to use your data to train models on our behalf beyond what's needed to provide the service, per their respective terms.

A note on safety

If a conversation you submit shows signs of possible physical danger, threats, coercion, stalking, or self-harm, the product responds with calm, serious guidance instead of playful commentary, and Roast Mode is disabled for that case. This is not a substitute for real support — if you're in danger or crisis, please reach out to a trusted person or a relevant professional or emergency service.