DDLC Safety Report
Content Warning

This site contains evidence of predatory and inappropriate behavior. Content may be disturbing.

DDLC Discord
Forensic Analysis

Public safety report exposing predatory behavior, grooming, and coded language within a large gaming community.

Top 50 Flagged Users

The most dangerous users identified by composite severity score across all messages.

Critical Risk Index

Filtered index of 77 users specifically flagged for Solicitation, Peer Collusion, or targeting minors.

Evidence Dossier

Raw context windows flagged at extreme severity (8+), preserving investigative context around the worst documented exchanges.

Methodology Summary

This report was generated using a two-stage AI forensic pipeline on a dataset of over 1.2GB of historical Discord logs — over 12 million messages analyzed. First, Llama Guard 4 12B filtered for overtly explicit or unsafe content. Then, Qwen3-14B-Instruct performed a sliding-window contextual analysis to identify insidious behaviors like grooming, predatory solicitation, and coded language (pedophilic dog-whistles).

Last Generated: 3/20/2026
Read full methodology →