Regional
On-ground in major regions. Understanding local context.
How moderation works
Moderation is a human process supported by technology. We combine AI detection, expert judgment, appeals, and continuous improvement to keep Zoiko safe and fair.
Read Community StandardsOur moderation teams
Trust & Safety
Addresses critical safety. Coordinates with law enforcement.
Appeals
Provides fresh, unbiased review of enforcement decisions.
Analytics
Tracks trends and effectiveness. Identifies gaps.
Policy
Develops and interprets standards. Ensures global consistency.
AI/ML
Builds detection systems. Identifies patterns at scale.
Moderators
600+ trained specialists reviewing nuanced cases with care.
The moderation process
1. Report & Detection
Content is flagged by users or detected by our automated systems. We receive 2.3 million reports monthly.
2. Initial Triage
Our systems categorize reports by likely violation type and priority. Critical issues are escalated immediately.
3. Automated Review
AI analyzes content against policies. For obvious violations, it recommends immediate action. For nuanced cases, it escalates to humans.
4. Human Review
Trained moderators review context, cultural factors, and account history. This is where nuance and judgment matter most.
5. Enforcement
We apply appropriate action: warning, content removal, account restriction, or ban. Decisions are consistent and documented.
6. Notification
Users are notified with clear reasoning. 95% of decisions are appealable, giving people a path to challenge us.
Principles that guide us
Context matters
We consider relationships, culture, sarcasm, and recent events. One phrase doesn't tell the full story.
Humans lead
AI assists but never decides alone. Every important decision involves human judgment and care.
Consistency
Similar violations get similar actions. We audit ourselves regularly for bias and fairness.
Transparency
Users know why we act. We publish quarterly reports and explain our decisions clearly.
Appeals work
2% of appeals overturn decisions. We take that feedback and improve our policies.
We evolve
Language, culture, and harm change. We continuously update policies and detection.
Challenges we face
⚡ Scale & speed
With 2.3M reports monthly, we must be fast. But speed without accuracy causes harm to innocent users.
🌍 Cultural diversity
What's normal in one country offends in another. Context from local teams is essential and irreplaceable.
🤖 AI limitations
AI struggles with sarcasm, metaphor, and niche communities. Human judgment fills the gap, but doesn't scale.
🎨 Gray areas
Many cases don't fit neatly into yes or no. We make judgment calls that some users will disagree with.
Help us build better safety
Your feedback—reports, appeals, and ideas—makes our moderation better. We listen, learn, and improve.