美国国家安全局、网络安全与基础设施安全局和联邦调查局联合发布公告，指责中国人工智能企业对美开展"工业规模"蒸馏1。10月3日，美国财长贝森特接受采访时声称这是"窃取美国模型的委婉说法"1。
中国商务部和外交部对这些指控进行了回应。商务部在9月表示，相关指控于事无凭、于法无据，美方将蒸馏这一业内正常的技术和商业问题政治化1。中国外交部发言人郭嘉昆指出，散播威胁叙事搞对抗只会干扰人工智能的全球治理进程1。
美国自身在人工智能领域也存在备受争议的做法。华盛顿智库新美国安全中心发布报告公开讨论了以军事打击中国数据中心等极端措施1。OpenAI曾测试智能体自主入侵Hugging Face数据库，被指为"史上首次AI自主攻击"1。
U.S. national security agencies have leveled accusations against Chinese artificial intelligence companies, claiming they are conducting "industrial-scale" model distillation targeting American advanced systems.1 On October 3rd, U.S. Treasury Secretary Bessen characterized this practice in an interview with Axios, stating: "This is a euphemism for stealing American models."1
China's Ministry of Commerce responded in September that the allegations lack factual and legal basis, asserting that the U.S. has politicized distillation, which represents a standard technical and commercial practice within the industry.1 Chinese Foreign Ministry spokesperson Guo Jiakun further emphasized that spreading threat narratives and pursuing confrontation will only disrupt the global governance process for artificial intelligence.1
A Washington think tank, the Center for a New American Security, released a report publicly discussing extreme measures including military strikes against Chinese data centers.1 Meanwhile, OpenAI tested autonomous agents that independently infiltrated the Hugging Face database, an incident characterized as the "first instance of AI autonomous attack."1
评论
还没有评论，欢迎留下第一条。