What is this page? (Zero-DOM Visual Perception & Canvas Vault Benchmark)
Standard privacy extensions only work by reading HTML text tags (<p>, <input>). When an AI browser agent encounters scanned documents, PDFs, Google Docs canvas, or Figma files, there are 0 DOM elements. This vault renders a classified flight manifest directly onto raw HTML5 canvas pixels. VEIL-X uses its on-device Tesseract WASM OCR and UltraFace ONNX visual engine to locate and redact sensitive data directly on the rendered canvas before external LLM transmission.