The US Retreats from the Global AI Governance Table
Article by Dr Megha Shrivastava and Sudhanshu Kumar
We have seen for years that Washington has tried to convince the world that it would set the guardrails for Artificial Intelligence (AI) and help the world navigate the uncertainties associated with AI and other emerging technologies. When Executive Order 14110 was signed by President Biden in October 2023, it wrapped AI governance in the language of “whole-of-government” urgency. Many experts widely saw it as the most comprehensive attempt to put certain rules around a runaway technology like AI. Even though it wasn’t perfect, it was a signal: a signal directing to the idea that America wanted to lead in AI development, and it was willing to constrain its own companies, at least on paper, to prove it.
At the recently held AI Impact Summit, where industry leaders, policymakers, and researchers convened across panels to discuss the future impact of AI, a common expectation was that the world must come together to formulate a coherent framework for the global governance of AI. The United States, long the self-appointed sheriff of the AI frontier, chose not to show up in the way it once did.
US representative Michael Krastios explicitly remarked, “We totally reject the global governance of AI”. This demonstrates a departure from its much-anticipated role in putting guardrails on the regulation of AI. Such an approach was visible even at the 2025 Paris AI summit, wherein the American position prioritised innovation over regulation.
What remains more concerning with the US departure is not that the sheriff is absent, but that it stands on an adversarial side. Additionally, when the US takes a backseat in the global governance of AI, it does not leave the world with a regulatory vacuum. The danger is not that norms disappear. It is that they get written by actors with very different values. It is more likely that the existing contestations between the states with different regulatory approaches will become more volatile.
One of the most concerning ramifications for the entire world is that the US retreat could create space for China to occupy the central role in the global governance of AI. China’s ongoing advocacy for AI governance apparently aims at maintaining a balance between innovation and ethical guardrails. It also seems to have had a real impact on AI governance conversations in countries in the Global South. China has secured the adoption of its own AI capacity-building resolution by the UN General Assembly, co-sponsored by more than 140 countries. China’s approach remains rather direct. It offers open-source models and deployment-ready technology. It is a compelling offer to cash-strapped, infrastructure-hungry nations. The Nigerian government has expressed support for Chinese AI governance initiatives. Indonesia has sought Chinese assistance in AI development for aquaculture and agriculture.
For India, this means the governance language that shapes the next decade of AI – how data is defined, how safety is measured, how responsible AI is framed may increasingly carry Beijing’s fingerprints. That is also a direct threat to India’s stated vision of being the alternate voice in this conversation.
Additionally, China’s low-key presence at the summit remains rather interesting. On one hand, it provides an opportunity for India to project itself as an advocate of the Global South cause. On the other hand, it means China is silently pursuing more impactful deliverables. Its rapidly rising influence through the flagship Digital Silk Road Initiative and the International Telecommunication Union (ITU) is well-known. On top of that, its impressive growth in open-source AI platforms (open-weights) has higher chances of gaining real traction for democratising the use of LLMs, against the West-led proprietary models like OpenAI’s ChatGPT.
While China’s prospects of gaining more influence in the AI governance space remain high, it is less likely that the global governance of AI will have a single sheriff for the foreseeable future. Thus, in this uncertain world of AI, India’s job is to help write new rules for global governance of AI that are fair rather than just powerful, to build coalitions that are principled rather than merely transactional, and to make the case that AI governance and AI ambition are not opposites.
To sum up, though the summit has ended, vacancy remains, and India has a genuine opportunity to emerge as a democratic alternative to China’s authoritarian vision. Given this, the achievements with the Delhi Declaration are not enough. India needs to push its vision into substance. First, beyond hosting summits, India must work on building institutions, which set standards, arbitrate disputes and give other Global South nations a table to deliberate on their more localised issues. Second, India must convert its “frugal AI” model into an exportable one in practice. Overall, creating an ecosystem for leading South-South cooperation on low-resource language AI. That is going to be India’s differentiator eventually.
Views are those of the Author(s)
