Chinese President Xi Jinping’s proposal for a BRICS open-source artificial intelligence community has put the spotlight on a question India has been trying to address: how can countries build advanced AI capabilities without becoming dependent on either US or Chinese technology ecosystems?
At the BRICS Summit in New Delhi, Xi proposed the creation of an open-source AI community that would promote cooperation among member countries on large language models and AI development. He said China would take the lead in establishing the community and building an open ecosystem for AI.
The proposal came months after India and France used the AI Impact Summit in New Delhi to advocate what they described as a “third way” for AI development—one that seeks to avoid excessive dependence on either Washington or Beijing.
Xi’s open-source AI proposal
Xi presented the BRICS AI community as one of five trade and technology initiatives at the summit’s outreach session.
He said China would support cooperation on developing and applying large language models and work towards an open AI ecosystem.
The proposal follows Beijing’s establishment of the World AI Cooperation Organisation, which brings together countries seeking alternatives to AI infrastructure dominated by US technology companies.
India has not yet publicly responded to Xi’s proposal.
What does open-weight AI mean?
The debate partly revolves around the distinction between open-weight and closed AI models.
In an open-weight system, the trained parameters of a model are made available for others to download, modify or host. Closed models, by contrast, keep their trained parameters private and provide access through services or application programming interfaces.
Models such as Meta’s Llama, DeepSeek and Alibaba’s Qwen have released open-weight versions.
However, open-weight does not necessarily mean fully open-source. Training data, development code and other parts of the model-building process can remain private.
Why India has an interest in open AI
Cost is one of the strongest arguments for open-weight AI. Countries and institutions can download and adapt existing models rather than building sophisticated foundation models entirely from scratch.
Open models can also offer greater technological autonomy. Indian universities, public institutions and companies could potentially fine-tune, host and audit models domestically without depending entirely on foreign AI APIs.
This is particularly relevant for India because of its linguistic diversity. The country’s scheduled languages and large multilingual population require AI systems capable of handling a wide range of languages and cultural contexts.
Open models could allow Indian researchers and institutions to adapt existing systems for local requirements instead of starting the entire development process from the ground up.
The risks of open AI
However, open-weight AI also presents significant challenges.
Once model weights are publicly released, they cannot easily be recalled. Researchers and policymakers have raised concerns that models could be modified to remove safety safeguards or used for harmful purposes.
Open systems can also make accountability more complicated. Responsibility may be spread among the original model developer, organisations that fine-tune the system, deployers and end users.
There is also a potential commercial challenge for India’s emerging AI companies. If increasingly capable open models become freely available, businesses and institutions may have less incentive to pay for domestically developed foundation models.
India’s ‘third way’ in AI
India and France used the AI Impact Summit in New Delhi earlier this year to advocate an approach positioned between the US and Chinese AI models of development.
The strategy emphasised three broad principles: strategic autonomy, shared democratic values and collaborative ecosystems.
Strategic autonomy means building sustainable domestic AI capacity while avoiding excessive dependence on a small number of dominant technology powers.
The second pillar focuses on safe, secure and transparent AI aligned with democratic values and human rights.
The third promotes open-access infrastructure, privacy-preserving data sharing and international research cooperation.
India also announced commitments focused on sharing anonymised AI-use insights and strengthening multilingual and contextual safety evaluations, particularly for the Global South.
How Xi’s proposal compares with India’s approach
Xi’s BRICS proposal closely matches one part of India’s strategy: the creation of collaborative AI infrastructure.
A BRICS ecosystem involving pooled computing resources, shared models, research cooperation and training could make advanced AI more accessible to developing countries.
But Beijing’s leadership creates a strategic dilemma for India.
India’s third-way approach is designed to prevent dependence on any single dominant technology power. Joining a Chinese-led AI ecosystem could potentially reduce dependence on US technology companies while creating a different form of dependence on Chinese technology and infrastructure.
The governance question is equally significant.
India has emphasised democratic values, human rights, safety and transparency in its AI approach, while China’s domestic AI framework operates under different regulatory and political principles.
That difference could become important if a BRICS AI ecosystem were to develop under significant Chinese leadership.
The choice facing New Delhi
Xi’s proposal therefore presents India with a complicated choice.
New Delhi could engage with the BRICS initiative because it offers access to shared infrastructure, AI research and potentially lower-cost technology. At the same time, India would have to consider whether participation in a Beijing-led ecosystem is compatible with its broader objective of maintaining strategic autonomy.
India’s original “third way” was intended to avoid choosing between Washington and Beijing.
Xi’s open-source AI proposal makes that strategic balancing act considerably more difficult.
