The G20 is moving forward on global AI governance—and the US risks being left out
Something notable happened in Johannesburg late last month, although it drew limited attention in Washington: Many of the world’s major economies signaled a growing alignment around how artificial intelligence (AI) and data should be approached—not primarily as instruments of geopolitical competition, but as vehicles for inclusive and sustainable development. The Group of Twenty (G20) leaders’ declaration, adopted despite uneven participation among several countries, reflects an important shift in how states are positioning themselves on AI governance. It offers a snapshot of an emerging global conversation that increasingly links AI to development goals and digital equity.
And the United States was not part of that moment.
The US delegation did not attend the Johannesburg summit and declined to join the declaration—a decision that stemmed in part from concerns about the host nation and broader disagreements with aspects of the process. And the United States is making AI a focus of the G20 summit it is hosting next year, an indication that it has not ruled out collaboration. Still, this year’s absence carried symbolic weight. It suggested a narrowing US appetite to engage multilaterally at a time when many governments are moving quickly to shape the rules and norms surrounding transformative technologies. Other capitals may reasonably interpret this as an opening: If Washington steps back from these discussions, others will step forward.
And many did.
The G20’s digital agenda this year went further than any previous summit in knitting together AI governance with sustainable development. What emerged from Johannesburg was a clear premise: AI is not just a commercial or security asset; it is a public good, one that must be governed collectively. Countries from South Africa to Brazil to India insisted that data governance, ethical guidelines, and inclusive digital infrastructure are not luxuries—they are developmental necessities.
What came out of Johannesburg wasn’t the usual tech-salon optimism or Western policy jargon. It was the voice of a world determined to stop the next wave of innovation from hard-wiring the injustices of the last. For example, the declaration insisted that AI must be “human-centered” and “development-oriented,” backed by trustworthy data governance—not just for privacy, but as the backbone of equitable AI. It called for digital public infrastructure and real capacity-building for countries long pushed to the margins of the digital economy. And it linked information integrity directly to democratic resilience. It aligned itself with the United Nations Educational, Scientific, and Cultural Organization’s (UNESCO’s) ethical AI framework and the United Nations’ resolutions on equitable technology.
Call it whatever you want: multilateralism, solidarity, or simple common sense. But the message was unambiguous. A broad group of the world’s largest economies came together to say that AI must serve humanity, not just the handful of companies and countries capable of building it.
The United States still has avenues to re-engage—not by dictating outcomes, but by participating as a genuine partner.
What makes the US absence so striking is that for decades it was the United States that championed precisely these kinds of conversations. US diplomats helped build the global internet governance system through international multilateral and multistakeholder fora, such as the Internet Governance Forum. American civil society was instrumental in pushing human rights into digital debates. American universities trained the researchers shaping AI ethics. Yet today, as major economies explore AI’s developmental dimensions, the United States is largely outside the room.
The US administration’s current approach to AI—marked by a preference for domestic industrial strategy and selective bilateral partnerships—reflects a hardening belief that multilateral governance is either futile or dangerous. In too many parts of Washington, there is a sense that global cooperation simply helps China; that multilateral institutions dilute US influence; and that if the United States leads on innovation, it doesn’t need to lead on rules.
This is a profound misreading of how power works in the digital age.
It is true, of course, that the United States remains the world’s AI frontrunner. Its companies build the most advanced models and its research institutions are unmatched—at least for the time being. But technological dominance without normative influence is brittle. Governance frameworks—data standards, safety norms, ethics principles—shape markets and behavior as much as silicon and compute. If the rest of the world agrees on a vision for AI grounded in development, inclusion, and human rights, and the United States is not part of that process, then Washington risks becoming a rule-taker rather than a rule-maker.
Some observers are already calling Johannesburg a win for China. There is some truth to that. Beijing has long argued that developing countries deserve a larger voice in global tech governance, with Chinese President Xi Jinping criticizing the idea of AI as a “game of rich nations,” a theme emphasized in Chinese state media coverage. And China’s investments in digital infrastructure across the Global South give it clear geopolitical advantages. With Washington absent, Beijing’s narrative—centered on equity, development, and multilateral dialogue—faces fewer obstacles.
But focusing solely on China misses the bigger story. Johannesburg was not a Chinese diplomatic triumph. It was a Global South diplomatic triumph. India, Brazil, South Africa, Indonesia, and others played central roles in shaping the digital agenda. They were not passive recipients of a Chinese vision; they were co-authors of something genuinely new: a multilateral AI framework that reflects their own developmental priorities. This agency was highlighted not only in the declaration but also in the reporting across the Global South, including South Africa’s official summit briefings.
None of this means the United States has been written off as an ally. But it does reflect a growing impatience among other states. Adopting the declaration without US support was not a rebuke; it was a recognition that global cooperation cannot wait for universal participation. A generation ago, such a move would have been unlikely. Today, it feels increasingly normal.
What should worry Washington most is that this shift comes at the precise moment when AI is beginning to reshape the global economy in ways as profound as industrialization. The International Monetary Fund estimates that AI could boost global growth by nearly a full percentage point, transforming labor markets, education, healthcare, and agriculture. It could concentrate power or democratize it. And the rules that govern these transformations are being written now.
To be clear, G20 declarations are nonbinding and often aspirational. Implementation will depend on infrastructure, innovation ecosystems, and the particular needs of member states. Still, the fact that the Johannesburg declaration so explicitly anchors AI within the sustainable development agenda—at a moment when US alignment with that agenda is often questioned—signals a meaningful shift in global positioning.
By staying home, the United States is making a bet that it can shape these norms later, through market dominance alone. But history suggests otherwise. Governance norms, once set, are sticky. They embed themselves in institutions, standards, and expectations. They shape how technologies are built and how they spread. And they rarely bend to accommodate a latecomer—even a powerful one.
It is telling that, while the world was forging a collective path in Johannesburg, Washington was charting a very different course at home with the launch of the Genesis Mission—an ambitious drive to harness AI for domestic innovation and national competitiveness. It’s a bold investment, but one that risks reinforcing a US approach to AI that is inward-looking and self-referential at the very moment the rest of the world is moving toward shared governance and collective benefit.
But retreat is not destiny. The United States still has avenues to re-engage—not by dictating outcomes, but by participating as a genuine partner. The G20 declaration did not emerge in a vacuum; it builds on existing foundations the United States helped create, including the Group of Seven’s Hiroshima AI principles and the Organisation for Economic Co-operation and Development’s (OECD’s) AI framework. Those earlier initiatives emphasized trustworthy, rights-based AI—but they lacked a deep developmental dimension. Johannesburg extends the trajectory, integrating ethical safeguards with the practical realities of inclusion and infrastructure.
If Washington wants to regain its normative footing, it can start by showing up. The upcoming India AI Impact Summit in February 2026—already gaining momentum as a convening of Global South digital priorities—offers a stage where the United States can listen rather than lecture, and even align itself with the developmental intent now shaping global AI norms. And with the United States set to host the G20 next year, it has a rare chance to reset: to bring the existing principles into conversation with the Johannesburg framework rather than treating them as competing visions.
The choice ahead is not between US power and multilateral governance. It is whether the United States can recognize that power now depends on multilateral governance—on shaping shared norms, not merely exporting products. Much of the world has signaled that AI must be human-centered, equitable, and globally accessible. The question is whether Washington is willing to take its seat—not at the head of the table, but at the table at all.
Konstantinos Komaitis, PhD, is a resident senior fellow with the Atlantic Council’s Democracy + Tech Initiative at the Digital Forensic Research Lab (DFRLab).
Further reading
Tue, Nov 25, 2025
After a lackluster G20 in South Africa, Trump can take the group ‘back to basics’ in 2026
New Atlanticist By Hung Tran
A 2026 agenda of economic and financial stability could refocus the group and make it more relevant.
Tue, Nov 25, 2025
Dispatch from South Africa: The G20’s center of gravity continues to shift
New Atlanticist By Alisha Chhangani
Emerging markets are building coalitions, designing financial tools, and articulating visions for multilateral reform with growing clarity and confidence.
Mon, Nov 24, 2025
Safety should be front and center in India’s vision for its AI Impact Summit
New Atlanticist By Trisha Ray
Despite the headline attention on impact, safety needs to be fundamental to India’s vision for artificial intelligence that engenders trust, inclusion, and empowerment.
Image: Brazil's President Luiz Inacio Lula da Silva and South Africa's President Cyril Ramaphosa attend a plenary session on the first day of the G20 Leaders' Summit at the Nasrec Expo Centre in Johannesburg, South Africa, November 22, 2025. REUTERS/Yves Herman
