Protecting browser-based ML pipelines while keeping authoritative inference server-side | AI BriefWire