Trust & Safety AI Translation
Amazon's Trust & Safety team needed to communicate how ML and GenAI-powered fraud detection systems process tens of billions of daily actions.
Challenge
Live robotics integration in broadcast environment requiring precise CAD sightline analysis and camera blocking
Legal/Trust & Safety required communication of enforcement actions (counterfeit ring busts) without revealing operational tactics that could aid bad actors
Explaining automated risk evaluation without implying unchecked surveillance, critical for stakeholder buy-in across European markets with strict privacy regulations
Date
2024
Client
Amazon
Role
Senior Creative Producer
Production
Riverview Systems Group, Inc
Scope
Broadcast Graphics
Visual Design
Creative Direction
Art Direction
Speaker Coaching
Video Production
Key Visuals
Storytelling support on stage and within the broadcast.
Solution
Human machine partnerships positioned LLMs as tools for investigator efficiency, not replacements.
Used the 6M → 700K bad actor metric as concrete evidence of system precision without revealing detection algorithms.
Motion graphics display decision velocity with all actions evaluated with human oversight checkpoints visible to avoid "decision mystery".
Simple overlay graphics display data over objects to illustrate vision.