You know how to learn. You want to point it at something that matters.
You are in a STEM or social science field. You have heard of AI safety but the landscape feels scattered. This program gives you a structured map — what the open problems are, where your skills fit, and who is working on what. Many Cohort 1 graduates came in with no prior safety background and left with a clear research direction.
Engineer · Practitioner
You build AI systems. You want to understand what makes them fail.
You are a software engineer, ML engineer, or data scientist. You ship things. The safety field can feel academic and hard to enter from industry. This is the on-ramp. You will build a working understanding of technical safety problems and connect with people doing this work at labs and research organisations.
Curious · Committed
You are not sure where you fit yet. You want to find out.
No prior AI safety background required. Basic programming familiarity is enough for the technical track. What matters is 5–7 hours per week and genuine motivation to engage with a hard, important problem.
Curriculum
10 weeks. Every week has a lab.
Week
Topic
Lab
W1
AI capabilities and the risk landscape
LLM API
W2
The alignment problem and specification gaming
Prompt safety
W3
Introduction to mechanistic interpretability
TransformerLens
W4
Robustness and reliability challenges
Model testing
W5
Adversarial attacks and basic defenses
Red-teaming
W6
Scalable oversight and AI control
Safety evals
W7–8
Open research problems — where to contribute
Research design
W9–10
Optional supervised project
Supervised
How it works
What to expect each week.
Each week has a live 2-hour session with an expert facilitator, a reading, and a hands-on lab. You are in a small pod of 6–8 for discussion. No recordings replacing real engagement.
Commitment5–7 hours per week. Optional project phase goes up to 10 hours.
FormatFully online. Live sessions, not pre-recorded. Small pods of 6–8.
Cohort size50 participants. Large enough for a real network. Small enough to know everyone.
CostNothing. Selection is on motivation and fit, not ability to pay.
From Cohort 1
In their own words.
"The facilitation model — not lecture-based — built real critical thinking rather than surface familiarity. I now apply this lens directly in my work on agentic automations and through my role at the UNESCO Women for Ethical AI South Asia Chapter.
Neha
PM/BA · UNESCO Women for Ethical AI, South Asia
"I was exploring AI Safety on my own. It was scattered. The cohort fixed that — structure, consistency, and people who actually took it seriously. That combination changed how I approached it.
Prasanna
Now at an impact-aligned startup
Where this leads
The program does not end at Week 10.
Phase 2
AI Safety Research Fellowship
High-performing graduates are eligible for our 12-week stipend-supported research program. Technical safety, governance policy, and AI policy engineering tracks. Output: papers, tooling, and policy memos for real institutions.
Community
Alumni network and weekly discussions
You join the AISIN alumni network — practitioners, researchers, and policymakers. Weekly AI safety discussions continue after your cohort ends. You stay connected to the field, not just the program.
Global field
ENAIS and AI Safety Atlas partnerships
AISIN partners with ENAIS and AI Safety Atlas. Graduates get access to global AI safety events, fellowship pipelines, and researcher introductions that are otherwise hard to reach.
Frequently asked
No. Basic programming familiarity is enough for the technical track. What matters is 5–7 hours per week and genuine motivation to engage with the problem.
No. Open globally. Cohort 1 included participants from Uganda, Southeast Asia, and Europe. The program is anchored in India's context but not limited to it.
You receive a certificate and join the AISIN alumni network. High-performing graduates are eligible to apply for Phase 2 — a stipend-supported 12-week research program producing real outputs.
AI Safety India Community, founded by Aditya Raj — SPAR Fellow, BlueDot Impact Alumni, Jailbreak Hackathon Top 30 (Grayswan), and Pathfinder Fellow. Partner organisations: ENAIS and AI Safety Atlas.
Apply — Cohort 2
The field needs more people.
Applications are open. 50 participants selected on motivation and fit. Both technical and governance tracks running together.