AI Governance at Crossroads: National priorities trump global cooperation

As AI capabilities advance rapidly, the governance gap becomes increasingly dangerous.

In the high-stakes arena of artificial intelligence (AI) governance, nations are positioning themselves like master chess players - each move calculated to advance national interests rather than foster global cooperation. In such a complicated world, it makes sense to do a comparative analysis of various AI governance approaches seen worldwide. The Takshashila Institution's inaugural State of AI Governance Report, 2024, in the writing of which the author was also involved, does exactly this. It reveals how this strategic maneuvering is happening. It shows how geopolitical considerations are increasingly overshadowing concerns about transparency, accountability, and individual rights, and makes some predictions about where this might be headed.

A Fragmented Global Landscape

What picture does a comparative analysis of AI governance approaches present? While lip service is paid to international cooperation at summits like the Paris AI Action Summit, major powers are primarily focused on securing competitive advantages. The United States has opted for a pro-market regulatory environment in line with its objective of prioritising geopolitical considerations to protect its technological edge. It has chosen to favour ease of doing business over governance concerns.

The European Union, consistent with its traditional approach, has adopted a comprehensive regulatory framework centred on transparency, individual rights, and ethics. However, this has not yet translated into regulatory leadership driving technological innovation, giving critics ammunition to challenge its approach. Meanwhile, China, true to its state-heavy model, continues its path of heavy state control – balancing national security imperatives with technological advancement.

India’s approach is particularly telling – a light-touch regulatory environment combined with investments in indigenous AI models. This mirrors developments across the Global South, where nations are striving to build sovereign capabilities while avoiding regulatory frameworks that might stifle innovation.

Corporate Self-Regulation: Substance or Theatre?

In this era of techno-nationalism and the race for technological sovereignty, governments will do what they deem necessary to maintain national interests and global standing. However, big tech has stepped into the governance void with self-regulatory initiatives. Yet the report raises important questions: can corporate self-regulation be meaningful without standardised reporting or robust external scrutiny?

While many companies are establishing principles and guardrails exceeding regulatory requirements, the effectiveness and sincerity of these measures remain uncertain. The recent controversy surrounding AI companies’ rollback of safety guardrails following the February 2025 AI Action Summit demonstrates the limitations of relying on corporate goodwill alone. Binding commitments and independent verification mechanisms are necessary to avoid corporate governance initiatives becoming mere window dressing.

The Limitations of Multi-Stakeholder Initiatives

The report's assessment of multi-stakeholder gatherings is illuminating. While high-profile events like AI summits have proliferated, concrete progress remains elusive. Countries have adopted isolationist approaches, reflecting the current environment of tech wars and deglobalisation.

While global governance may inevitably return – as game theory suggests – at present there is more discord and divergence than cooperation. The USA and EU’s refusal to sign the declaration on inclusive AI at the February 2025 Paris AI Action Summit exemplifies this challenge. These forums are essentially paying lip service to governance and ethics. Without concrete, binding commitments backed universally, their outcomes are negligible. This is reminiscent of the evolution of global climate change action – a parallel that provides little comfort.

The AI governance landscape reflects deeper tensions in international relations. When national interests diverge, cooperative frameworks struggle to gain traction – as seen in the contrasting approaches to open-source AI models: China and the EU view them as pathways to strategic autonomy, while other nations are sceptical, concerned about uncontrolled proliferation.

Predictions: Geopolitics Will Continue to Overshadow Ethics

Looking ahead, Takshashila’s predictions suggest these trends will intensify, at least in the short term. Regulatory triggers based on compute thresholds will become obsolete, as smaller models grow more efficient. This exposes a fundamental flaw in current regulatory approaches that focus on inputs rather than capabilities. Innovations around efficiency and scale will make this shift inevitable.

Nations are expected to increase investments in sovereign cloud infrastructure – driven more by geopolitics than technological needs. State-led AI governance will continue prioritising innovation over transparency, accountability, and societal well-being in the near future.

The prediction that US chip restrictions on China will not escalate further – due to innovations like DeepSeek, which reduce the need for the latest chips – highlights the adaptive nature of the AI innovation ecosystem. Similarly, the forecast that EU regulations may lead some companies to withhold AI products from European markets suggests that even comprehensive regulatory frameworks face implementation challenges. It is predicted that China and the EU will continue to promote open-source and open-weight models as a route to strategic autonomy and technological leadership. A harbinger of this could be the continued open-weight/open-source status of players like DeepSeek and Mistral.

The Path Forward: Beyond Zero-Sum Thinking

If fragmented, nationally focused governance continues, we risk squandering AI’s immense benefits while failing to mitigate its risks. Without meaningful international cooperation, regulatory arbitrage will weaken even the best-designed governance systems.

What is needed is a recognition that effective AI governance requires both competitive and cooperative elements. Nations can pursue strategic advantage while collaborating on shared challenges – such as ensuring AI systems are safe, secure, and aligned with human values.

The emergence of “Chief Responsible AI Officer” roles within companies deploying AI solutions at scale, as predicted by the report, may be a small step in the right direction. However, without parallel developments in robust international governance mechanisms, corporate efforts will remain insufficient.

As AI capabilities advance rapidly, the governance gap becomes increasingly dangerous. Takshashila’s analysis serves as a wake-up call: we must move beyond the current chess game of national interests towards global cooperation that acknowledges both competitive realities and our shared stake in ensuring AI benefits humanity.

The question is how do we overcome geopolitical rivalries before technology outpaces our capacity for thoughtful oversight.