Transparency Report

Trust is earned through openness. We publish aggregated safety and moderation statistics — no PII, ever.

H1 2026 (Preview)

Published 2026-06-29Next report: December 2026

Download report (Markdown)Safety Center

Content Moderation

Aggregated moderation statistics across HushSafe-powered platforms. No personally identifiable information is included.

2.4M

Interactions reviewed

99.8%

Classification accuracy

<2%

Appeal rate

118ms

Median review latency

Safety Actions

Actions taken to protect users across the ecosystem.

184K

Harmful content prevented

12.4K

Accounts actioned (policy violations)

8.1%

Appeals upheld

Privacy & Data

How we handle data requests and user rights under DPDP and GDPR.

Data access requests fulfilled

Deletion requests fulfilled

User data sold

Model Cards & Bias Testing

HushSafe text classifier v2.1 — quarterly bias testing across language, dialect, and demographic slices. Full model cards published in the Developer Hub.

Languages supported

Bias test slices

June 2026

Last audit

Questions about this report? Contact trust@matrioshkalabs.com. File a grievance via our Grievance Officer.