01
Research
Institutional AI safety, AI sovereignty, deployment governance, machine unlearning, interpretability and context-specific evaluation.
NEW DELHI · RESEARCH · DIPLOMACY · CAPACITY
Institutional AI safety for an interdependent world.
Project AWARE is a public-interest initiative advancing institutional AI safety and practical AI governance for the Global South, middle powers, smaller states, and other non-frontier countries.
We focus on the institutions, deployment routes and international relationships through which powerful AI systems enter public and economic life.
01 / WHY WE EXIST
AI safety does not end at the model.
It is shaped by the institutions that procure, deploy, audit, govern and contest AI systems.
For countries that do not control frontier model development, the central question is not only how a model behaves. It is also where governing power can be exercised once that model enters a domestic market, public institution or regulated sector.
02 / OUR WORK
Research that can move into institutions.
Project AWARE combines policy research, technical inquiry, diplomatic engagement and researcher development.
02
Diplomacy
Structured dialogue across policy, diplomatic, legal, technical and industry communities on the interests of non-frontier states.
03
Capacity
Research programmes that help emerging practitioners translate AI safety and governance questions into rigorous public-interest work.
03 / SELECTED RESEARCH
Recent papers and live work.
Governing Foreign Generative AI Through Deployment Routes in the Global South
Where policy leverage may sit for states that do not own frontier model providers, with attention to cross-border API, cloud-mediated enterprise and public-sector deployment.
Read the paper ↗
Beyond Universal Metrics
A Global South-centered framework for ethical AI development.
Open paper ↗
Unlearning at Scale
Implementing the right to be forgotten in large language models.
Open paper ↗
Markovian Circuit Tracing for Transformer State Dynamics
A controlled benchmark and evaluation pipeline for studying coarse state-transition structure in transformer activations.
Open paper ↗
Live
Do Frontier AI Models Prefer Elite Names?
An ongoing study of Chilean elite-coded surname signals and whether those representations produce decision bias.
View live study ↗
Regulate or enable?
Mixed signals on AI infrastructure cause for concern.
Read commentary ↗
04 / A RESEARCH LENS
Power may sit in the route, not only the model.
Foreign AI systems reach domestic institutions through a chain of infrastructure, contracts, data flows and deployment decisions. Those routes can create practical points of governance.
01
Containment
Restrict where, how or for what purpose a system may be used.
02
Bargaining
Use market access, procurement and infrastructure conditions to demand safeguards.
03
Structural
Build durable influence through standards, public capacity and cooperation.
05 / CONVENING
From principles to institutional practice.
Project AWARE convenes focused discussions around the practical machinery of AI governance.
INDIA AI IMPACT SUMMIT 2026
From Guidelines to Ground
Institutional AI Safety for India and the Global South
Date16 February 2026
VenueBharat Mandapam, New Delhi
FormatOperational roundtable
The discussion examined procurement, audit, oversight, incident response and remedy in high-adoption contexts where institutional capacity is uneven.
Organised with
The Dialogue
Cyril Shroff Centre for AI, Law and Regulation
View event record ↗
06 / CURRENT PROGRAMME
AI GOVERNANCE RESEARCH COHORT · 2026
Building a country-by-country account of governing foreign AI.
The inaugural cohort brings together contributors across law, public policy, international relations, AI governance, economics and technical systems.
Its central work is a comparative paper series examining how selected states can build stronger leverage over foreign AI systems used in their markets, public institutions and regulated sectors.
Follow programme updates ↗07 / LEADERSHIP
A small institution built around serious work.
Project AWARE was founded in 2024 and operates from the National Capital Region, India.
Founded by Abdullah X, Project AWARE is led across research, diplomacy and capacity building with Fiza Siddiqui, Director.
Founder and Chair
Abdullah X
Director
Fiza Siddiqui
08 / CONTACT
The countries most dependent on AI systems must not become the countries least able to govern them.
For research collaboration, institutional partnerships, diplomatic engagement and programme enquiries.
contact@zephara.ai