As AI systems expand globally, the challenges of global AI governance have become harder to ignore. Nations are struggling with issues like equitable representation, multilingual accessibility, and striking the correct balance between supervision and innovation. Global digital rights activists emphasized the US’s waning support and the changing priorities of tech at the RightsCon summit in Taiwan.
In the meantime, recent initiatives such as Europe’s “sovereign AI” plans and India’s push for web browsers demonstrate a shifting balance of power in the digital space. JD Raimondi and Ruban Phukan advocate for more intelligent solutions, such as localized AI regulations and flexible language models.
Who Will Lead in Global AI Governance?
Large tech companies are gradually moving away from helping smaller user groups, particularly those that don’t speak English. Governments and innovators are responding accordingly. Zoho Corporation, Team PING, and Team Ajna are the three local projects chosen by MeitY’s browser challenge in India to address local digital needs.
The Euro Stack and other EU plans for digital sovereignty represent a change in strategy. Furthermore, European leaders are reconsidering their reliance on platforms based in the United States. This modification highlights the necessity of global language models that support local languages. The goal of Mozilla’s volunteer-based dataset project and tools from Indian startup Shhor AI is to lessen bias in AI systems. The system has been trained primarily in American English.
Is Global AI Governance Becoming Geopolitical Now?
The global AI governance conversation turned geopolitical at the recent Paris AI Summit. French President Emmanuel Macron pitched France as Europe’s AI leader. However, Sumedh Nadendla pointed out that the “large” models of today might appear “small” in the future. Concerns about who controls language models are growing rapidly. These worries are rising as DeepSeek-V3 now boasts 671 billion parameters.
Civil society organizations expressed concern at RightsCon about the United States’ decision to reduce funding for digital rights. Meanwhile, the Netherlands and India are developing their infrastructure, such as India Stack.
Open-source small models (SLMs) are advocated by Ruban Phukan for inclusive growth. He also highlights that laws governing AI must promote openness without stifling creativity. The cost of training and implementing smaller models is lower. Additionally, they function well in offline, mobile-first environments, which are typical in developing nations.
Smaller Models Could Change AI Development Path
To enhance global AI governance, smarter AI regulation is essential. A recent study comparing AI policies worldwide found significant differences. China employs a command-control strategy, whereas the UK takes an ethics-driven stance. According to the study, regulations should be stricter in situations of competition.
Automated moderation will continue to overlook harmful content in susceptible areas in the absence of improved tools. However, Raimondi believes that SLMs, which stand for smaller, simpler models, could level the playing field.
Ultimately, inclusive datasets, balanced oversight, and multilingual access are essential for the success of global AI governance. To ensure AI serves all languages and cultures, governments must work with innovators and provide support.
Nations Must Act for Tech Equality Now
The stakes are higher than ever since AI has emerged as a major part of international competition. Sovereign tech plans can foster resilience, but digital isolation must be avoided. Furthermore, as language models advance and smaller models become more advanced, inclusivity should remain a top priority. Therefore, global AI governance must be a cooperative effort based on respect for diversity, local context, and transparency.
About The Author
