# Why Translating Scanned Legal Documents Is Still Broken in 2026 (And How We Are Fixing It) > Source: > Published: 2026-06-24 10:39:49+00:00 If you have ever tried translating a scanned property deed, court order, birth certificate, or government document from Hindi to English, you know how frustrating the process can be. The document is usually: Most AI translation tools fail at the very first step. Traditional OCR struggles to extract text accurately from Indian documents. Generic AI translators may produce grammatically correct sentences but often miss legal terminology, context, formatting, and names. This becomes a serious issue because in legal documents, a single mistranslated phrase can completely alter the meaning. We discovered this problem firsthand while working with hundreds of Indian legal and government documents. A land record is not the same as a court judgment. A sale deed is not the same as an affidavit. A birth certificate issued in Rajasthan looks very different from one issued in Tamil Nadu. Yet most translation systems treat them all as ordinary text. That is why we started building **Lekhak**. Lekhak is an AI-powered legal document translation platform designed specifically for scanned and image-based documents. Instead of performing simple word-to-word translation, the platform first understands the document structure, extracts text from images, identifies legal entities, preserves formatting, and then performs context-aware translation. The goal is simple: **Make legal and government document translation accurate, fast, and accessible for everyone.** Over the coming weeks, I will share: If you work in legal tech, immigration, compliance, courts, or AI, I would love to hear your experiences. What is the most difficult document you have ever had to translate?