Case Study: InSpace improves chat safety with TensorFlow.js toxicity filtering

InSpace A new video conferencing platform that uses TensorFlow.js for toxicity filters in chat

InSpace, a virtual learning and communication platform, sought to foster a safe and collaborative environment by preventing toxic messages within its chat feature. The challenge was to identify and filter out inappropriate comments contextually, moving beyond a simple profanity blocklist, without storing or processing user data on a server for privacy.

To solve this, InSpace implemented a pre-trained toxicity detection model from TensorFlow.js. The TensorFlow solution runs entirely within the user's browser using a Web Worker to ensure the filtering process is non-blocking and does not impact performance. By integrating the TensorFlow model, InSpace can now automatically warn users before sending toxic messages, successfully creating a safer and more moderated chat experience for educators and students.

InSpace

Narine Hall

Chief Executive Officer