返回首页
Source: CCTV News
According to a June 14 report from CCTV News, the U.S. government has recently resolved to ban all foreign entities and foreign nationals residing within the United States from accessing two of Anthropic’s latest artificial intelligence models. U.S. authorities announced the decision on the 13th, disclosing that a discussion between Amazon and U.S. officials prompted the final ruling. Amazon alleged that critical vulnerabilities exist in Anthropic’s newest AI models.
Sources familiar with the matter stated that Amazon CEO Andy Jassy held talks with U.S. officials including Treasury Secretary Bessent. During the meeting, Jassy shared that Amazon researchers had crafted a series of prompts to manipulate one of Anthropic’s latest models into releasing restricted information that could be leveraged to facilitate cyberattacks.
The U.S. government subsequently held meetings to deliberate countermeasures and verify Amazon’s claims. Federal officials ultimately concluded that blocking foreign entities and foreign users from accessing the models was the most straightforward solution.
In a statement released on the 12th, Anthropic acknowledged the government’s authority to block unsafe deployments, yet argued such restrictions ought to be enforced through transparent, fair, legally mandated procedures grounded in technical evidence — standards which the current government action fails to meet.
The company issued an apology to its customers and stated it is working to restore access as soon as possible. Anthropic further contended that the vulnerabilities flagged by Amazon are relatively basic flaws also present in other publicly available models, and maintained these loopholes cannot enable unauthorized hacking of electronic devices.
Notably, Amazon serves as Anthropic’s core strategic investor, cloud service provider and chip supplier for its data centers. Amazon also deploys Anthropic’s advanced models within its own software vulnerability detection business.
Back in March this year, Anthropic was added to the federal government’s supply chain risk blacklist after refusing the U.S. Department of Defense’s demand for unrestricted use of its AI technology. The firm immediately filed a lawsuit against the Department of Defense in response.
Analysts pointed out that the U.S. President recently signed an executive order tightening oversight of advanced AI models, and reports previously surfaced that the U.S. government is considering taking equity stakes in AI companies. Taken together, these moves demonstrate the current administration’s push to strengthen state control over the artificial intelligence industry.
美国政府下禁令：禁止外国主体使用Anthropic最新AI模型
来源：央视新闻
据央视新闻6月14日报道，近日，美国政府决定禁止所有外国主体和境内外国人使用安特罗匹克公司（Anthropic）两个最新型人工智能模型。美国方面13日消息称，美国亚马逊公司与美国官员之间的一次沟通促使美国政府最终做出了这一决定。亚马逊方面指出，安特罗匹克公司的最新人工智能模型存在漏洞。
据知情人士消息，亚马逊首席执行官安迪·贾西在与包括美国财政部长贝森特在内的美国官员沟通时称，亚马逊的研究人员曾使用一系列提示词，诱导安特罗匹克公司的一个最新人工智能模型提供本应受到限制、且可用于协助网络攻击的信息。
随后美国政府方面开会商议对策并验证亚马逊方面的说法。最终美国政府官员认为阻止外国主体和外国人士访问该模型是最直接的方法。
安特罗匹克公司在12日发布的声明中表示，政府有权阻止不安全的部署，但应当通过透明、公平且基于技术事实的法定程序行使，而政府当前的做法并不符合以上原则。
声明还向客户致歉，并称正在努力尽快恢复访问权限。安特罗匹克公司还认为，亚马逊指出的这类漏洞相对来说比较基础，其他公开可用的模型也能发现这些漏洞，并声称这些漏洞不会构成“对电子设备的非授权破解”。
值得一提的是，亚马逊公司是安特罗匹克公司的核心战略投资者、云服务提供商，也是后者数据中心的芯片提供商，同时亚马逊还将安特罗匹克先进的模型部署于自身的软件漏洞识别业务中。
今年3月，安特罗匹克公司因拒绝美国防部无限制使用该公司人工智能技术的要求，被列入联邦政府供应链风险黑名单。该企业随即对美国防部提起诉讼。
分析人士认为，近期美国总统签署关于强化人工智能模型审查的行政令。此前还有消息称，美国政府考虑入股人工智能企业。这都说明，本届政府正在试图加强对人工智能行业的控制力度。
U.S. Government Issues Ban: Foreign Entities Barred From Accessing Anthropic’s Newest AI Models
Source: CCTV News
According to a June 14 report from CCTV News, the U.S. government has recently resolved to ban all foreign entities and foreign nationals residing within the United States from accessing two of Anthropic’s latest artificial intelligence models. U.S. authorities announced the decision on the 13th, disclosing that a discussion between Amazon and U.S. officials prompted the final ruling. Amazon alleged that critical vulnerabilities exist in Anthropic’s newest AI models.
Sources familiar with the matter stated that Amazon CEO Andy Jassy held talks with U.S. officials including Treasury Secretary Bessent. During the meeting, Jassy shared that Amazon researchers had crafted a series of prompts to manipulate one of Anthropic’s latest models into releasing restricted information that could be leveraged to facilitate cyberattacks.
The U.S. government subsequently held meetings to deliberate countermeasures and verify Amazon’s claims. Federal officials ultimately concluded that blocking foreign entities and foreign users from accessing the models was the most straightforward solution.
In a statement released on the 12th, Anthropic acknowledged the government’s authority to block unsafe deployments, yet argued such restrictions ought to be enforced through transparent, fair, legally mandated procedures grounded in technical evidence — standards which the current government action fails to meet.
The company issued an apology to its customers and stated it is working to restore access as soon as possible. Anthropic further contended that the vulnerabilities flagged by Amazon are relatively basic flaws also present in other publicly available models, and maintained these loopholes cannot enable unauthorized hacking of electronic devices.
Notably, Amazon serves as Anthropic’s core strategic investor, cloud service provider and chip supplier for its data centers. Amazon also deploys Anthropic’s advanced models within its own software vulnerability detection business.
Back in March this year, Anthropic was added to the federal government’s supply chain risk blacklist after refusing the U.S. Department of Defense’s demand for unrestricted use of its AI technology. The firm immediately filed a lawsuit against the Department of Defense in response.
Analysts pointed out that the U.S. President recently signed an executive order tightening oversight of advanced AI models, and reports previously surfaced that the U.S. government is considering taking equity stakes in AI companies. Taken together, these moves demonstrate the current administration’s push to strengthen state control over the artificial intelligence industry.
美国政府下禁令：禁止外国主体使用Anthropic最新AI模型
来源：央视新闻
据央视新闻6月14日报道，近日，美国政府决定禁止所有外国主体和境内外国人使用安特罗匹克公司（Anthropic）两个最新型人工智能模型。美国方面13日消息称，美国亚马逊公司与美国官员之间的一次沟通促使美国政府最终做出了这一决定。亚马逊方面指出，安特罗匹克公司的最新人工智能模型存在漏洞。
据知情人士消息，亚马逊首席执行官安迪·贾西在与包括美国财政部长贝森特在内的美国官员沟通时称，亚马逊的研究人员曾使用一系列提示词，诱导安特罗匹克公司的一个最新人工智能模型提供本应受到限制、且可用于协助网络攻击的信息。
随后美国政府方面开会商议对策并验证亚马逊方面的说法。最终美国政府官员认为阻止外国主体和外国人士访问该模型是最直接的方法。
安特罗匹克公司在12日发布的声明中表示，政府有权阻止不安全的部署，但应当通过透明、公平且基于技术事实的法定程序行使，而政府当前的做法并不符合以上原则。
声明还向客户致歉，并称正在努力尽快恢复访问权限。安特罗匹克公司还认为，亚马逊指出的这类漏洞相对来说比较基础，其他公开可用的模型也能发现这些漏洞，并声称这些漏洞不会构成“对电子设备的非授权破解”。
值得一提的是，亚马逊公司是安特罗匹克公司的核心战略投资者、云服务提供商，也是后者数据中心的芯片提供商，同时亚马逊还将安特罗匹克先进的模型部署于自身的软件漏洞识别业务中。
今年3月，安特罗匹克公司因拒绝美国防部无限制使用该公司人工智能技术的要求，被列入联邦政府供应链风险黑名单。该企业随即对美国防部提起诉讼。
分析人士认为，近期美国总统签署关于强化人工智能模型审查的行政令。此前还有消息称，美国政府考虑入股人工智能企业。这都说明，本届政府正在试图加强对人工智能行业的控制力度。