앤스로픽, 中 AI 경쟁 경고 "지금 잡지 않으면 늦다"

Anthropic frames AI competition with China as a now-or-never moment for Washington Maximilian Schreiner View the LinkedIn Profile of Maximilian Schreiner May 15, 2026 Anthropic Key Points In a policy paper, Anthropic warns that the US must defend its lead in AI computing power against China. Despite US export controls, Chinese labs stay competitive by smuggling chips and systematically copying American model capabilities through fake accounts. Anthropic sees two scenarios for 2028: either the US closes these loopholes, or China catches up and pushes authoritarian AI norms worldwide. Ask about this article… Search In a policy paper, Anthropic lays out two scenarios for 2028: either the US locks in its compute lead over China, or authoritarian regimes set the rules for the AI era. The timing is no coincidence. Anthropic has published an extensive policy paper that frames the AI competition between the US and China as a geopolitical make-or-break question. The core argument: democracies need to defend their lead in computing power, or they risk an AI era shaped by authoritarian regimes. Compute, meaning access to advanced AI chips, is the critical bottleneck, Anthropic argues. Thanks to US export controls and innovation at companies like Nvidia, TSMC, and ASML, democratic nations hold a commanding lead in hardware. According to an analysis cited in the paper, Huawei will reach only four percent of Nvidia's aggregate compute capacity by 2026, dropping to just two percent by 2027. Ad Still, Chinese AI labs remain close to the frontier. Anthropic points to two reasons: chip smuggling and access to US compute through foreign data centers on one hand, and systematic distillation attacks on the other. In these attacks, Chinese labs use thousands of fake accounts to scrape the outputs of American frontier models and replicate their capabilities. Back in February, Anthropic accused Deepseek, Moonshot, and Minimax of generating more than 16 million Claude interactions through roughly 24,000 fraudulent accounts. OpenAI, Google, and the joint Frontier Model Forum have condemned the practice as well . Ad DEC_D_Incontent-1 What happens if the US closes the loopholes, and what happens if it doesn't Anthropic sketches two futures. In the first scenario, US policymakers close the loopholes in chip exports and cloud access, crack down on distillation through legislation, and promote the global export of American AI infrastructure. The result: a 12- to 24-month lead in model intelligence, US AI as the backbone of the global economy, and AI norms shaped by democracies. In the second scenario, the loopholes stay open. Chinese labs reach near-parity, Beijing scales AI-powered surveillance and repression, and Huawei data centers gain global market share with cheaper but capable enough models. Anthropic warns that a neck-and-neck race like this would also erode safety standards: only 3 of 13 top Chinese labs have published safety evaluations so far, and DeepSeek's R1-0528 fulfilled 94 percent of malicious requests under common jailbreaking techniques, according to the Center for AI Standards and Innovation . Ad The paper dropped on May 14, 2026, right in the middle of Trump's visit to China and while Congress is actively debating AI export controls, chip licenses, and cloud access. Anthropic is planting its flag firmly on the restrictive, security-focused side. The pitch to policymakers is simple: if the US closes the loopholes now, it can lock in a lasting AI advantage through 2028. If it doesn't, American technology will help China's AI ecosystem catch up. Ad DEC_D_Incontent-2 AI News Without the Hype – Curated by Humans Subscribe to THE DECODER for ad-free reading, a weekly AI newsletter, our exclusive "AI Radar" frontier report six times a year, full archive access, and access to our comment section. Subscribe now Source: Anthropic