Insights for the people keeping platforms safe.
What's working in trust and safety: posts from our team, case studies from platforms using Stream, and short guides on specific harms and how to stop them.
From the blog.
View all postsWhat we see across 8B+ moderation requests a month, and how we build for it.
Stream Moderation Benchmarks: 1,500+ Requests per Second.
How the moderation pipeline holds up at peak load.
- Marketplace Moderation Statistics: What 550M+ Checks Show.Where harmful content shows up most on online marketplaces.
- Inside Stream AI Moderation for Live Video & Audio.How frames and live captions are scored and flagged while the stream runs.
- The Moderation Metrics Every Trust & Safety Team Should Track.The numbers to report on and what they tell you.
- Transparent Content Moderation: Mistakes, Appeals and Statements of Reasons.How to handle appeals and explain decisions to users.
Harm guides: one harm at a time.
How each harm shows up, why filters miss it, and what to set up to stop it.
- Evasion
Moderation evasion: algospeak and obfuscation
Coded slang, character tricks and format switching that slip past filters, and how layered detection catches them.
- Spam
Bot detection with behavioral signals
Bots that post spam and scam links while blending in, and how to spot them from their behavior over time.
- Scams
Marketplace scams and abuse
Fake listings, impersonation and off-platform payment requests, and how to stop them without slowing down sellers.
Stream moderation FAQs.
Trust and safety is the work of keeping users safe on a platform: writing policies, detecting harmful content and behavior, and acting on it. That covers scams, harassment, child safety, spam and abuse in text, images, video and audio.
Each guide covers one type of harm, such as filter evasion and algospeak, spam bots and marketplace scams. It explains how the harm shows up, why basic filters miss it and what to set up to stop it.
No. The guides explain how each harm works, what to look for and how to set up protections on any platform, whatever moderation tools you use.
New blog posts and guides are added as we see new harms and attack patterns across the platforms we moderate.
Yes. Use the form at the bottom of this page to tell us what you're seeing on your platform, and we'll consider it for a future guide.
Dealing with a harm we haven't covered?
Tell us what you're seeing. We'll walk you through how to set it up in Stream.