AI Use Policy

Indias.com uses AI to accelerate research, drafting and multilingual expansion at national scale. But large language models can hallucinate — invent facts, sources, statistics and people who do not exist. This policy documents the guardrails we run so AI helps our editors without misleading our readers.

(a) First-draft reference articles from a structured research packet compiled from official sources; (b) Summarising long official documents; (c) Translation of already-approved English/Hindi content into 55 supported languages; (d) Voice narration ("Listen to this page"); (e) Suggested internal links and image alt-text; (f) The "Ask India" grounded Q&A assistant.

(1) Inventing contact details, addresses, phone numbers, coordinates, accreditations, opening hours, prices or ratings; (2) Writing about living public officials without human editorial review; (3) Predicting elections, court outcomes or market movements; (4) Generating medical, legal, immigration or tax advice; (5) Creating photographs of real people or real events; (6) Publishing anything without a human editor's approval.

Every article states in its footer whether the current version was AI-drafted-then-edited, AI-translated-from-a-reviewed-English-original, or fully human-written. The "Ask India" assistant cites the specific Indias.com entity or external source that informed every factual answer.

We use AI illustrations only as a themed decorative fallback when no licensed photograph is available (e.g., topic tiles for regions where Unsplash has no match). We never publish AI-generated images depicting real people, real events, or real locations that would mislead a reader.

Text: OpenAI GPT-4o / GPT-4-class models via the Emergent LLM key. Images: Unsplash photographs (Creative Commons, credited), plus in-house SVG illustrations. Voice: browser Web Speech API (free, on-device). Model choices are reviewed quarterly.