The events of 2025 made clear that the question is no longer whether artificial intelligence (AI) will reshape the global order, but how quickly—and at what cost.
Throughout the year, technological breakthroughs from both the United States and China ratcheted up the competition for AI dominance between the superpowers. Countries and companies raced to build vast data centers and energy infrastructure to support AI development and use. The scramble for cutting-edge chips pushed Nvidia’s valuation past five trillion dollars—the first company to reach that milestone—even as concerns mounted over circular financing and the question of how much the AI boom is founded on hype versus reality. Meanwhile, policymakers grappled with the balance between safety, security, and innovation and how to manage possible labor disruptions on the horizon.
As 2026 begins, rapid AI integration threatens to inject even more unpredictability into an already fragmented global order. Below, experts from the Technology Programs share their perspectives on what to expect from AI around the globe in the year ahead.
Click to jump to an expert prediction:
Emerson Brooking: AI poisoning goes mainstream
Tess deBlanc-Knowles: The US pushes AI tech exports to counter China
Konstantinos Komaitis: AI governance turns global
Ryan Pan: The US-China AI race intensifies in a multipolar world
Esteban Ponce de León: AI challenges human judgment
Trisha Ray: Countries go all in on ‘sovereign AI’
Mark Scott: The battle of the AI stacks escalates
Kenton Thibaut: China doubles down on AI-powered influence operations
AI poisoning goes mainstream
Russia’s Pravda network of websites has published millions of articles targeting more than eighty countries. These sites launder and amplify content from Russian state media, seeking to legitimize Russian military aggression while casting doubt on Western support for Ukraine. Most of these articles will never be viewed by a human. Instead, they seem intended to target the web crawlers that scour the internet for training data to feed to insatiable AI models.
And the strategy is working. Last year, the ’s Digital Forensic Research Lab (DFRLab) and CheckFirst demonstrated how mass-produced Pravda articles were cited in Wikipedia, X Community Notes, and responses from major chatbots. Parallel research by Anthropic and the United Kingdom’s AI Safety Institute has shown how trace amounts of faulty data can effectively “poison” even very large models. People increasingly turn to AI systems to understand current events. If an AI model’s knowledge has been altered by sources intended to deceive, then the users’ will be, too.
In 2026, the issue of AI poisoning will break into the mainstream. Because of a roughly two-year lag in AI training data (many AI models are still waiting for the results of the 2024 US presidential election, for instance), these AI-targeted propaganda campaigns are about to start manifesting more often. And because one cannot reliably audit what’s inside a deployed AI model, the result will be a staggering research and policy challenge.
Digital policy experts, including the DFRLab, have spent a decade learning to identify, explain, and expose online disinformation where people can see it. This is online disinformation where they can’t.
—Emerson Brooking is the director of strategy and a resident senior fellow with the ’s Digital Forensic Research Lab.
The US pushes AI tech exports to counter China
In 2026, the United States will double down on exporting the US tech stack as the cornerstone of its international AI strategy. In December 2025, US President Donald Trump set the tone with his decision to allow Nvidia to export its advanced H200 chips to China, a clear endorsement of the view that the United States wins when the world builds and deploys AI using US technology.
Published days before the Nvidia decision, the Trump administration’s National Security Strategy makes this explicit: “We want to ensure that US technology and US standards—particularly in AI, biotech, and quantum computing—drive the world forward.” This framing echoes the AI Action Plan the administration released in July 2025, which stated that the “United States must meet global demand for AI by exporting its full AI technology stack,” warning that a failure to do so would be an “unforced error.”
In 2026, expect to see the United States sign more AI-focused partnerships like those forged with Saudi Arabia and the United Arab Emirates in 2025, alongside efforts to counter China’s growing influence in emerging markets. But as the United States makes this push, China holds some key advantages. Its lead in open-source AI models and focus on applied AI could prove to be the winning formula for capturing global market share with free models and deployment-ready technologies.
—Tess deBlanc-Knowles is the senior director of Technology Programs.
AI governance turns global
In 2026, AI governance enters its first truly global phase with the United Nations–backed Global Dialogue on AI Governance and Independent International Scientific Panel on AI. For the first time, nearly all states have a forum to debate AI’s risks, norms, and coordination mechanisms, signaling that AI has crossed into the realm of shared global concern.
Yet this ambition unfolds amid acute geopolitical tension: The European Union pushes a rights- and risk-based regulatory model, while the United States favors voluntary standards to preserve innovation and security flexibility. For its part, China promotes inclusive cooperation while defending state control over data and AI deployment. Smaller and developing states gain a voice but remain structurally dependent on the major powers that control the bulk of AI talent, capital, and computing power.
The result is a fragile, uneven global framework. States converge on scientific assessments, transparency norms, and voluntary principles, but they avoid binding limits on high-risk AI uses such as autonomous weapons, mass surveillance, or information manipulation. Coordination emerges, but the core strategic competition remains unresolved, producing a governance architecture that manages risks at the margins while leaving rival models largely intact.
By the end of 2026, the Global Dialogue will likely have made AI governance global in form but geopolitical in substance—a first test of whether international cooperation can meaningfully shape the future of AI or merely coexist alongside competing national strategies. This juncture offers states an opportunity to demonstrate leadership by strengthening institutional capabilities and collaborative mechanisms, fostering a global AI governance framework that is more coherent, equitable, and universally engaged.
—Konstantinos Komaitis is a resident senior fellow with the ’s Democracy + Tech Initiative.
The US-China AI race intensifies in a multipolar world
The year ahead will see an even fiercer competition over AI dominance between the world’s two largest powers—the United States and China—while middle powers gradually close the gap in the race. China’s DeepSeek started off this year with a research paper on a new AI training method to efficiently scale foundational models and reduce costs. This publication comes almost exactly a year after the headline-making paper it released in January 2025, which was followed by the launch of DeekSeek-R1. The timing of this year’s new publication signals that the company will launch new models and continue shaping the world’s AI industry this year.
In 2026, expect China to double down on its open-source AI strategy to influence the world’s AI infrastructure. (Several major US tech companies are already using Chinese large language models in their applications.) The United States and China may also engage in further trade retaliation in the AI supply chains in light of recent developments in Venezuela, from which Chinese companies had gained access to rare earth minerals crucial to developing the AI stack. The Trump administration’s recent claims regarding Colombia, from which China also sources rare earth elements, could make Latin America the next technology battleground between the two powers.
But what about powers beyond the United States and China? In 2026, look for Europe to increase its AI defense investments even more than it did in 2025. Middle powers, notably India, will see their AI capability greatly improved this year, as US tech giants have recently pledged billions in investments in India’s AI capabilities.
The AI race in 2026 will still be defined by a multipolar order. Nevertheless, the United States and China will continue to yield the greatest influence.
—Ryan Pan is a program assistant with the ’s GeoTech Center.
AI challenges human judgment
In 2026, human–AI interaction will likely challenge human judgment and identity more deeply than in any year to date. This is not only because AI models are demonstrating increasingly complex capabilities, but also because AI-generated content can be so emotionally charged in today’s polarized information environment.
Online sources and social media have shown how polarization can be deliberately targeted, and the use of AI to generate fabricated or distorted content adds a new layer to how social and political events are interpreted. AI content is reshaping the dynamics of both manipulation and what could be described as a “misinformation game,” in which techniques such as the deployment of AI slop and the memeification of events are used to mock adversaries and amplify key propaganda narratives. For example, in June 2025, amid the Israel-Iran escalation, AI became the new face of propaganda. This included graphic and sensational AI-generated fake content, such as fabricated missile strikes, military hardware, religious and national symbols, and memes. But it also included more sophisticated fabrications of CCTV footage that became increasingly difficult to debunk.
In the first days of 2026, as the Trump administration captured Venezuelan strongman Nicolás Maduro, the use of AI to generate media content increased drastically. While much of this content was humorous or satirical in nature, it nonetheless illustrates emerging usage patterns, as playful AI-generated media can still shape perceptions of power and blur the line between satire, manipulation, and propaganda. Whether fabricated content aims to provoke humor or confusion, human judgment will face new challenges in the year ahead.
This challenge to human judgment and identity extends beyond misinformation. In 2026, the AI landscape may begin to show early signs of benchmark saturation, in which models converge at near-maximum scores on established capability tests, collapsing the measurable differences between them. This matters for the information environment because the same logic applies: If distinguishing real from fabricated content becomes difficult, then so too does distinguishing what humans uniquely contribute from what AI can replicate. The implications extend to professional identity and how to understand individual value and competence.
—Esteban Ponce de León is a resident fellow with the Digital Forensic Research Lab.
Countries go all in on ‘sovereign AI’
There are unprecedented amounts of capital flowing in to meet the anticipated demand for AI. Last year, for instance, kicked off with Trump’s announcement of Stargate, with the aim of investing $500 billion in AI infrastructure over five years. The principle driving this trend is straightforward: Countries think they must control AI before it controls them. Consequently, there was a wave of sovereign AI announcements in 2024 and 2025.
That momentum will only grow in 2026, starting with the launch of India’s sovereign large language model at the AI Impact Summit in February. Nations are seeking sovereign AI to strengthen their domestic economies, protect national security, mitigate geopolitical shocks, and reflect national values. However, there’s a catch: Not every country can, or should, try to build every part of the AI stack on its own. Trying to recreate from scratch everything from data centers to models is expensive, redundant, and impractical. Nations will need to choose what to build, what to buy, and where partnerships make more sense than going solo.
—Trisha Ray is an associate director and resident fellow with the GeoTech Center.
The battle of the AI stacks escalates
As AI becomes more central to countries’ economic prospects, national policymakers will likely seek to impose greater control over critical digital infrastructure. This infrastructure includes compute power, cloud storage, microchips, and regulation, and it is central to how emerging AI technology will develop in 2026. For the world’s largest digital powers—the United States, the European Union, and China—the push to control this infrastructure will likely evolve into a battle of the “AI stacks”—increasingly opposing approaches to how such core digital AI-enabling infrastructure functions at home and abroad.
The White House’s AI Action Plan, published in July 2025, made it the stated policy of the federal government to export the US stack to third-party countries, including via potential funding support from the US Department of Commerce for other governments to purchase offerings from the likes of Microsoft, OpenAI, and Nvidia. The European Commission has earmarked billions of euros for so-called AI gigafactories, or high-performance computing infrastructure, from Estonia to Spain, while national leaders also vocally called for a “Euro stack.” The Chinese Communist Party is urging local firms to forgo Western AI know-how and rely instead on domestic alternatives from companies such as Alibaba or Huawei.
The rest of the world will have to navigate these increasingly rivalrous approaches to AI infrastructure at a time when all countries seek greater control of so-called digital public infrastructure—that is, the underlying hardware and, increasingly, software needed to power complex AI systems. How these different AI stacks interact with each other will be critical to how the technology develops over the next twelve months.
—Mark Scott is a senior resident fellow with the ’s Democracy + Tech Initiative.
China doubles down on AI-powered influence operations
In 2026, the People’s Republic of China’s (PRC’s) AI-enabled disinformation efforts are likely to intensify in scale, persistence, and technical sophistication, particularly those targeting Taiwan. PRC actors are already using AI-generated audio, video, and text, distributed through networks of fake accounts and contracted private firms, to conduct “cognitive warfare” campaigns aimed at shaping political perceptions and voter behavior. These campaigns prioritize volume, localization, and algorithmic exploitation, and they are increasingly designed to be continuous rather than episodic. As AI-generated content is blended with human-curated messaging and commercial infrastructure, PRC-linked operations will become harder to detect and attribute, reflecting a shift toward more deniable, adaptive, and professionalized influence operations.
At the same time, Beijing is expected to pair these activities with defensive diplomatic messaging that rejects allegations of PRC-linked disinformation or cyber operations and reframes such claims as politically motivated attacks. This pattern reinforces a broader hybrid strategy in which AI-enabled influence operations, cyber activity, and diplomatic signaling are tightly integrated. In 2026, PRC disinformation campaigns are likely to focus less on overt propaganda and more on shaping narratives around crises and cyber incidents, contesting blame, eroding trust in attribution, and influencing strategic decision-making outcomes.
—Kenton Thibaut is a senior resident fellow with the Democracy + Tech Initiative.
