U.S. officials claim Chinese AI companies stole American technology
美國官員聲稱中國人工智慧公司竊取了美國技術
更新於: 2026年9月10日 上午02:45
On September 8, 2026, U.S. national security agencies, including the FBI and NSA, issued a formal warning regarding the theft of American artificial intelligence technology.
2026年9月8日，美國國家安全機構，包括聯邦調查局（FBI）與國家安全局（NSA），針對美國人工智慧（AI）技術遭竊發布正式警告。
The government alleges that six Chinese companies—including DeepSeek and Alibaba—have engaged in industrial-scale intellectual property theft.
美國政府聲稱，六家中國企業（包括深度求索〔ㄕㄣˋㄔㄡˊ〕與阿里巴巴）涉嫌進行工業規模的智慧財產權竊取。
According to the advisory, these firms used a technique called 'knowledge distillation' to illicitly copy the capabilities of frontier AI models developed by companies like OpenAI, Google, and Anthropic.
根據該建議書，這些公司利用名為「知識蒸餾」的技術，非法複製由 OpenAI、Google 與 Anthropic 等公司開發的前沿 AI 模型功能。
By using covert proxy accounts and fraudulent methods, these companies allegedly bypassed security guardrails to train their own 'student' models on data stolen from U.S. 'teacher' models.
透過使用隱蔽的代理帳戶與詐欺手段，這些公司據稱繞過了安全防護機制，利用從美國「教師」模型中竊取的資料來訓練自身的「學生」模型。
This activity allows firms to save significantly on development costs and research time.
此活動讓這些企業大幅節省了開發成本與研究時間。
U.S. officials worry that this stolen technology could bolster China's military and cyber capabilities.
美國官員擔心這些遭竊的技術可能增強中國的軍事與網路能力。
This issue represents a significant diplomatic challenge ahead of a summit between President Donald Trump and President Xi Jinping, highlighting the ongoing global competition for AI dominance.
此議題在唐納·川普（Trump）總統與習近平主席舉行高峰會前，構成了一項重大外交挑戰，凸顯出全球在 AI 主導權上的持續競爭。
