IndiaAI developers are prioritising bias checks in foundational LLMs while promoting open access AI tools under the global AI Commons initiative.
Entities developing large language models (LLMs) under the IndiaAI Mission have been instructed to make bias mitigation a top priority before deployment. Given India’s diverse social landscape, officials emphasised that government-supported AI models must avoid producing insensitive or discriminatory outcomes related to caste, gender, food practices, regional, linguistic, ethnic, or religious differences.
A MeitY official said,”Sensitive connotations linked with caste, gender, food practices, regional and linguistic stereotypes, as well as ethnic and religious differences have to be handled with utmost care. We want Indian models to be inclusive, and not discriminatory or based on historical biases. As a result, all the under-construction LLMs have been told to integrate stringent stress testing into their framework.”
Another official added, “The sovereign LLMs are a crucial milestone of our AI journey, which should bring the country together. The focus should not shift from this, towards bad actors or miscreants trying to orchestrate a frenzy by feeding compromising prompts to AI. We need to be careful since machine learning tools process data on a massive scale; even small biases in the original training data can lead to widespread discriminatory outcomes.”
In October, IndiaAI invited expressions of interest for Stress Testing Tools (STT) to evaluate models under adversarial inputs, data drift, and distributional shifts, ensuring fairness, privacy, and accountability.
Startups and consortia developing indigenous AI include Param-1, BharatGen, Tech Mahindra’s Indus LLM, Sarvam AI’s open source Sarvam M, and Soket AI’s DHRITH.
The government is advancing the AI Commons initiative, a global, open-access framework featuring ethical AI certification, anonymisation, and stress testing, encouraging worldwide collaboration. This approach promotes transparent, inclusive, and community-driven AI development, positioning India as a leader in ethical open source AI.
