AI-Powered Censorship Solution for Telekom Malaysia, Tier-1 Malaysian National IPTV Platform
Project Overview
Developed an AI-enabled, cloud-based content censorship system for the client's IP Media Center operations.
The solution assists content compliance by automatically detecting nudity, foul language, sensitive content, and gore imagery, enabling AI-assisted pre-screening to accelerate manual review processes while maintaining human oversight.
The solution is deployed on AWS and integrates with ingest, frame analysis, and encoding workflows (via SKYLARK), enabling scalable and high-volume media operations.
Our Contribution & Software Capabilities
- Achieved 85% detection accuracy, significantly reducing manual "eye-balling" effort;
- Reduced operational load while maintaining human-in-the-loop governance;
- Contributed to AI-driven scene and audio analysis, enabling automated pre-screening prior to human review;
- Assisted in QA, model validation, and tuning to improve detection reliability and operational confidence.