Anthropic warns US has until 2028 to secure AI dominance over China or risk authoritarian control

Anthropic Issues Urgent Call for Stronger Controls in US-China AI Race

Anthropic has released a comprehensive policy paper warning that the United States has a narrow window until 2028 to secure its position in the global AI competition against China1

1

. The San Francisco-based AI developer argues that advanced AI development will deliver transformational economic and societal impacts in the coming years, and the outcome depends critically on where the most capable systems are built first1

1

. The company's timing is notable, as the policy paper emerged during President Trump's official trip to China, deliberately inserting itself into ongoing discussions about technological sovereignty4

4

.

Two Contrasting Futures for AI Governance by 2028

The Claude maker outlines two starkly different scenarios for what the world could look like in 2028, when it expects transformative AI systems to have emerged1

1

. In the first scenario, America has successfully defended its compute advantage, allowing democracies to set the rules and norms around AI development1

1

. This would enable the US to guide AI toward transparency, accountability and ethical use, reinforcing democratic values3

3

. The second scenario presents a darker picture: China overtakes the US, leading to AI norms and rules shaped by authoritarian regimes, with the best models enabling automated repression at scale1

1

. Anthropic warns that Artificial General Intelligence (AGI) could arrive by 2028, describing a future with a country of geniuses in data centers where AI models can accelerate scientific discovery and even AI research itself5

5

.

China's AI Ambitions Fueled by Workarounds and Innovation

Despite existing chip export controls, Chinese AI labs have built models that come close to those developed in America, according to the policy paper1

1

. Anthropic attributes this to China's talent pool and their ability to exploit loopholes in export restrictions, along with distillation attacks that illicitly extract innovations from American companies1

1

. The company claims that export-controlled Nvidia silicon is either smuggled into the country or accessed via offshore datacenters, while Chinese firms employ distillation attacks on top US frontier models to stay competitive4

4

. According to one China-based cybersecurity analyst, the preview of Claude Mythos sent serious shockwaves, with the analyst describing it as facing a fully automatic Gatling gun while China is still sharpening swords4

4

.

Semiconductor Export Controls as Critical Battleground

Anthropic analyzed Nvidia and Huawei's roadmaps for manufacturing silicon and concluded that Huawei will only be able to produce roughly 4% of Nvidia's total AI compute capability in 2026 and just 2% in 20274

4

. This means China cannot bridge the AI compute gap with Nvidia on its own, and if chip export controls are tightened, the US will have roughly 11 times more AI compute than China4

4

. The paper argues that advanced semiconductor chips are the most important ingredient in frontier AI development, and currently democracies, especially the US, dominate in producing the latest AI chips4

4

. The company called for tighter chip export controls and restrictions on overseas compute access to preserve the US advantage5

5

.

Geopolitical Implications and the Stakes of Authoritarian AI

The policy paper emphasizes that advanced artificial intelligence will soon become one of the world's most important geopolitical tools, shaping military power, cyber capabilities, economic growth and global influence2

2

. In China, AI is already being used for mass surveillance, automated repression and military advancements, with AI-driven facial recognition systems deployed to track individuals in real-time and predictive policing algorithms targeting potential dissenters3

3

. The company specifically warned that authoritarian governments could potentially use powerful AI systems for mass surveillance, cyber operations and social control5

5

. If China overtakes the US in AI leadership, these practices could become international norms, threatening democratic freedoms worldwide3

3

.

Criticism and Alternative Perspectives on Export Controls

The policy paper has drawn criticism for showing blinkered thinking by implying that China can only advance by riding on America's coattails and is incapable of innovating, despite the shockwaves generated by the release of the DeepSeek R1 model early in 2025, believed to be on par with the best US models1

1

. Many experts argue against export control, suggesting DeepSeek as evidence that it actually helps China innovate around the hurdles faster4

4

. Jeffrey Ding, a prominent US-based expert on China's AI capabilities, points out how AI dominance among nations shouldn't just be based on innovation but also on diffusion of technology to people at scale4

4

. Another problem with the plan is that many countries, especially in Europe, view both American and Chinese AI supremacy as a threat to democracy, with a concerted push for digital sovereignty to minimize reliance on US technology1

1

.

US Policy Response and Path Forward

Anthropic can draw little comfort from the Trump administration, which has a constantly shifting attitude toward the US-China AI competition1

1

. Export controls were reportedly not high on the agenda during the President's trip to Beijing, and it was reported that the US has now cleared around 10 Chinese firms to buy Nvidia's second-most powerful AI chip, the H2001

1

. To maintain technological dominance, Anthropic argues the US must tighten export controls, disrupt distillation attacks and accelerate AI adoption in democratic nations, which could allow a 12 to 24-month advantage over Chinese rivals3

3

5

5

. Failure to act will allow China to export its AI infrastructure and models to the world, especially the Global South, with Huawei and Alibaba growing to take a greater chunk of the global AI market share4

4

.