Stream’s AI Moderation acts as the first line of defense in keeping your community safe. Instead of expecting human moderators to scan every single message, post, or image, the AI reviews all content in real time and flags items that may violate your community standards.
For moderators, this means that you don’t see everything users post; you see what the AI has already identified as potentially harmful. Your role is to review those flagged items, confirm whether the AI’s decision was correct, and take the appropriate action.
What AI Detects
Stream’s AI can flag harmful content across multiple categories, including:
- Harassment and Bullying: Insults, threats, or targeted abuse.
- Self-Harm: Expressions of suicidal thoughts or encouragement of self-harm.
- Sexual Content & Exploitation: Sexual harassment, adult content, or child safety violations.
- Hate Speech & Terrorism: Attacks on protected groups or extremist content.
- Spam & Scams: Fraudulent promotions, phishing attempts, or repeated unwanted content.
- Media Violations: Unsafe images or videos containing nudity, violence, or disturbing content.
Each of these harms may look different depending on your community, but moderators will always see the AI’s label for why a piece of content was flagged.
How AI Decisions Work
When content is posted, the AI instantly analyzes it for context, tone, and intent. Unlike simple keyword filters, it looks at conversation history to tell the difference between:
- A joke between friends vs. targeted harassment.
- A news article mentioning violence vs. a threat of violence.
If the AI detects a risk, it assigns:
- A label (e.g., Harassment, Spam, Self-Harm).
- A severity level (text) or confidence score (media).
- A default action (e.g., Flag for review, Block, Shadowblock).
That information is what you’ll see in your queue and use to guide your decision-making.
Why This Matters for Moderators
The AI doesn’t replace you, it supports you. By filtering out safe content and surfacing only the riskiest cases, AI helps you:
- Save time by avoiding endless scrolling.
- Focus on the edge cases that require human judgment.
- Apply actions more consistently across users and channels.
In short, AI moderation makes the work manageable, but moderators make the final call.
Now that you understand what Stream’s AI Moderation detects and why content shows up in your queue, let’s explore the dashboard itself, your command center for reviewing flagged items, accessing queues, and using moderation tools effectively.