GUZOLIVESafety center
← Creator StudioDK
Protected by policy + people

Keep the room worth returning to.

AI catches patterns at the edge, moderators make the hard calls, and paid interactions never skip the safety queue.

● AI service healthy
Queue depth042 high confidence · 2 held
Auto-filter rate98.4%last 60 minutes
Avg. decision420msp95 · all languages
Muted today17temporary · appealable
Live queue · Battle of Adwa

Review before it reaches the room

Newest first · paid and free comments follow the same policy.

ViewerMessageSignalAction
TK
Tesfaye K.New account · 12 sec ago
“This round is absolutely electric 🔥”
Held · 0.62
MN
Mimi N.Verified · 28 sec ago
“Can we get a replay of the last exchange?”
Review · 0.71
AB
Account 4831Repeated text · 1 min ago
Repeated message detected across 4 submissions.
Spam · 0.96
SA
Samuel A.Trusted viewer · 2 min ago
“The crowd is louder than the commentary!”
Held · 0.54
Policy controls

Event guardrails

Changes apply to this event run.

Draft synced
AI moderationNormalize + classify every message
Broadcast-safe filterStricter threshold for OBS output
Slow modeOne text comment every 10 seconds
Paid commentsHold for approval before spotlight
AI thresholds

Confidence window

Ambiguous content stays held.

Fail closed
More reviewMore automation
StrictOpen
Language coverage

Ready for the room

4 active
EnglishአማርኛTransliterationCode-switching

Production adapter returns category scores, severity, confidence, detected language, model version, and policy version.