ModerateHatespeech - Free toxicity detection for content moderation via AI

Hey ProductHunters! We're ModerateHatespeech, a nonprofit project helping online communities fight and study hate speech. Online moderators are crucial to the health, growth, and inclusivity of online communities, whether that's Facebook Groups, subreddits, Discord servers, forums, or Slack channels. Technology and anonymity have made it drastically easier to troll and submit verbally abusive content online, thereby increasing the overall toxicity of a community. In turn, that has been shown through research to dissuade potential users from participating in discussions, driving down user engagement. Moderators have an essential but frequently overlooked role -- for many sites which receive hundreds of thousands of comments a day, scanning through all content submitted is impossible, and thus moderators primarily rely on user reports. Yet, our own findings who that online ~1 in 10 pieces of rule-breaking, hateful/insulative content is actually reported. And, it can take hours for such content to be reported, reviewed, and removed, allowing it to incite further clashes and escalations. Our completely free API is trained on a massive corpus of hundreds of thousands of comments in all text styles. With our technology and integrations (including for WordPress, Reddit, Discord, and more) you can: - Automatically remove hateful content above a certain threshold as soon as it is submitted - Flag (but not remove) hateful content for review - Analyze the historic behavior of a user, to understand if they have a pattern of rule-breaking content Since we're powered by AI, we can contextually discern between casual teasing and result insults, as well as understand different uses of works (ie, "F**k") providing significantly higher accuracy and detection rates than keyword/phrase filters. Currently, we help moderate communities with tens of millions of members, reducing toxicity by ~ 85% just through our integration. Because of our high-powered, highly optimized approach, our models have significantly few false positives when in comes to content moderation, ensuring anything that is flagged or removed is almost certainly malicious. For non-moderators, our API can aid in research and analysis of hate speech and toxicity on a large scale online, identify high-toxicity profiles or sites, and more. As a nonprofit, we're committed to transparency, open access, and gaining a better understanding of the propagation of toxicity online. You can read more about some of our projects here: https://moderatehatespeech.com/r... We do not log or store any data submitted through our API, so the discussions of your communities are completely confidential. We'd appreciate any thoughts or feedback -- interested users can sign up for an API key directly on our site, or email us if you need any help at all. Welton & the ModerateHatespeech team

Congrats on the launch! I'm building a comment moderation tool that hides toxic language, profanity and spam on Facebook & Instagram posts and ads. For this tool I'm using Perspective API to score toxicity. Does your service use Perspective API too? Or would it classify our users comments differently compared to their toxicity score?

Great idea! The demo is giving me error though

@weltonwang we're planning to give this a try for Goosly when we introduce activity comments for public roadmaps. I mentioned ModerateHatespeech in my live stream last evening:

Congrats with the launch!

does it work on instagram and youtube?