[LT-04]
The RTL Compliance Gap
Arabic and Hebrew present a systematic PII detection failure for organizations using tools built primarily for left-to-right Latin-script languages. The problem is not merely directional. Right-to-left scripts require different tokenization, different segmentation logic, and different entity boundary detection than LTR approaches. Standard NER systems trained on English data apply LTR segmentation assumptions that produce incorrect entity boundaries in Arabic and Hebr...