AI Safety India

AI Safety India (aisafetyindia.com) is a national Indian community initiative whose mission is to "bring together students, researchers to work on the most important challenge of our time — aligning advanced AI systems with human values." Founded in 2025, the group is led by Aditya Prasad (Founder), Aditya Raj (Partnerships Lead), and Sireesha Chavali (Advisor), and runs India's first structured AI-safety cohort programme: a three-phase pipeline — Fundamentals of Safe AI, Research Fellowship, and Research Paper — designed to move participants from zero AI-safety background through independent research and publication, offered free of charge. It is the corpus's first entry covering the AI-safety community-building layer in South Asia.

Founding and motivation

The founding motivation is documented in Aditya Raj's founding-story series: despite a large number of Indians active in the global AI-safety field, "there are so many people in AI Safety from India, but nothing significant is happening for India." Raj consulted over 40 field builders for feedback before launch, built the organisation's website using his design background, and produced a vision document outlining the group's intended trajectory. The group operates nationally and primarily online, with no single city base.

Supporting and collaborating organisations include ARENA, EA Hungary, ENAIS (European Network of AI Safety), Stanford Center for AI Safety, AI Safety Asia, and AI Plans — a partnership footprint that situates AI Safety India in both the global EA-adjacent AI-safety community-building network and the specialist technical-safety skills pipeline.

Three-phase programme

AI Safety India structures its work around a three-phase curriculum. Phase 1 — Fundamentals of Safe AI is a 10-week free online cohort running approximately 6–10 hours weekly, covering AI safety foundations (alignment, risks, ethics, governance), technical topics (interpretability and robustness), and hands-on projects modelled after the BlueDot Impact and Atlas programmes. Sessions run in small groups of 5–10 participants — a format that keeps cohort density high enough for substantive peer exchange without losing the guided-cohort structure. The inaugural cohort opened applications on April 28, 2025, received over 100 applications, and accepted approximately 40 participants; the programme ran June through mid-August 2025. Phase 1 is designed for students and early-career professionals with basic Python knowledge but no prior AI-safety background.

Phase 2 — Research Fellowship and Phase 3 — Research Paper advance participants beyond the fundamentals cohort into guided independent research and through to a completed publication — a pipeline modelled on the progression from BlueDot's foundational courses to original technical contributions.

Place in the movement

AI Safety India is the corpus's first entry for an AI-safety community-building group in South Asia, filling the geographic gap the Synthesizer identified (zero India or South Asia entries across the local-groups layer). It sits in the insider-community pattern alongside Berlin AI Safety, AI Safety 東京, AI Safety South Africa, and the Singapore AI Safety Hub (SASH): EA-adjacent community groups whose central activity is building the local practitioner pipeline for technical AI-safety work. What distinguishes AI Safety India from those counterparts is the cohort-and-curriculum structure — where Berlin and Tokyo are built around regular monthly meetup-and-talk formats, AI Safety India leads with a structured education programme that moves participants from fundamentals to research across three phases. This reflects the field-building challenge in India: with many technically capable Indian students interested in AI-safety work but no existing local-group infrastructure to onramp them, a curriculum-first approach addresses the gap more directly than a meetup-first model.

A co-existing India AI-safety community layer — Safety & Alignment Research India (SAAR), a Discord community rated "Very Active" on aisafety.com — runs alongside AI Safety India as a peer discussion space for Indian alignment researchers and newcomers; the two operate complementarily rather than as competitors, addressing the same population through different formats (structured curriculum vs. open research-community channel).