ThinkPatternGet the app
Story
TECHNOLOGY · MAY 15, 2026

Anthropic Warns U.S. Faces 24-Month AI Race Window Amid Trump-Xi Summit

Anthropic urged the U.S. to tighten chip export controls within 24 months to maintain its AI lead over China, as Trump and Xi discussed AI guardrails in Beijing.

Anthropic published a policy paper warning that the United States has 12 to 24 months to secure a decisive lead in frontier artificial intelligence before China reaches near-parity by 2028. The company identified Chinese firms DeepSeek, MiniMax, and Moonshot AI as illicitly using Claude outputs through distillation attacks to train smaller competing models. Anthropic also cited chip smuggling and offshore datacenters as methods China uses to close the compute gap. The company urged the U.S. government to tighten export controls on advanced semiconductors, increase enforcement budgets, and block distillation attacks, arguing that authoritarian AI leadership under the Communist Party would enable repression at unprecedented scale.

The warning coincided with a two-day summit in Beijing between President Donald Trump and Chinese President Xi Jinping, where the leaders discussed AI safety protocols and standard guardrails. Trump stated the two explored working together on guardrails for AI. The summit also addressed Nvidia H200 chip exports: while the U.S. had eased restrictions to allow shipments, China has not approved purchases, preferring to develop domestic semiconductor alternatives. U.S. Trade Representative Jamieson Greer called the approval a sovereign decision for China, and Commerce Secretary Howard Lutnick confirmed Beijing has blocked its companies from buying the licensed chips.

The timing highlighted a policy tension. Anthropic pressed for stricter containment, while reports indicated the Trump administration may soften some export restrictions. Treasury Secretary Scott Bessent confirmed discussions with China on AI governance frameworks. Meanwhile, industry experts including Alvin Wang Graylin of the Stanford Institute for Human-centered Artificial Intelligence criticized Anthropic's arms-race framing as irresponsible and self-serving. Beijing's recent order for Meta to unwind its acquisition of AI startup Manus further underscored the escalating technology rivalry between the two nations.


Reported across 121 outlets
Actors
Donald TrumpFederal government of the United StatesGovernment of ChinaNvidiaAnthropicXi Jinping

Keep reading in the app

The full story and every source, free in the app.

Download on the App StoreComing soonGoogle Play