字号·· | 护眼
亚洲新闻台

马来西亚要求Anthropic解释在影响力行动中对人工智能的使用Malaysia wants Anthropic to explain AI use in influence campaign

点「原文对照」整页切到原文,或双击某段只看那段的原文。

这张插图于2026年10月19日拍摄,展示了Anthropic公司的标志和一个键盘。(资料图:路透社/达多·鲁维奇)An Anthropic logo and a keyboard in this illustration taken on Oct 19, 2026. (File photo: Reuters/Dado Ruvic)

吉隆坡:政府表示,马来西亚正向美国人工智能公司Anthropic索取信息,因为该公司涉嫌在其针对马来西亚选民的政治影响力行动中使用了其技术。

KUALA LUMPUR: Malaysia is seeking information from US artificial intelligence firm Anthropic over the alleged use of its technology in a political influence operation targeting Malaysian voters the government has said.

Anthropic在9月发布的一份报告中表示,该公司发现了一起疑似利用人工智能驱动的行动,旨在扰乱这个东南亚国家的选举。

Anthropic said in a report issued in September it had uncovered a suspected AI-powered campaign meant to disrupt elections in the Southeast Asian nation.

报告指出一个被认为总部设在土耳其的“海外实体”,该实体滥用了Anthropic的Claude人工智能模型,在社交媒体平台X上生成了大约1,000个虚假账户,并建立了一个合成新闻网站。

It pointed to an "overseas entity", believed to have been based in Turkey, that abused Anthropic's Claude AI model to generate around 1,000 fake accounts on social media platform X and create a synthetic news site.

马来西亚通讯部周二(11月17日)在一份提交给议会的书面声明中,针对一名议员关于此事的提问回应称:“根据该报告,Anthropic指称Claude技术被一个海外实体用于开发该行动的几个组成部分,包括一个目标锁定系统、一个虚假社交媒体账户网络、虚假新闻网站及相关组件。”

"Based on the report, Anthropic alleged that Claude technology was used by an overseas entity to develop several components of the operation, including a targeting system, a network of fake social media accounts, fake news sites and related components," the Malaysian communications ministry told parliament in a written statement on Tuesday (Nov 17), responding to a lawmaker's questions about the matter.

该部表示,马来西亚通信及多媒体委员会正“就该报告向Anthropic寻求更多信息和澄清”,随后将“确定适当的后续行动”。

The ministry said the Malaysian Communications and Multimedia Commission was "seeking further information and clarification from Anthropic regarding the report", and would then "determine appropriate follow-up actions".

根据Anthropic的报告,该实体“将该平台的访问权限作为一种付费的影响力即服务能力进行出售”,并将该基础设施营销为“军用级、人工智能驱动的实时政治行动生态系统”。

According to Anthropic's report, the entity "sold access to the platform as a paid influence-as-a-service capability", marketing the infrastructure as "military-grade, AI-driven, real-time political operations ecosystem".

报告称:“该行动涉及各行为体利用通过Claude构建的平台,使用他们已录入的人口普查和选举数据,按选区对马来西亚选民进行画像和锁定。”

"The operation involved the actors leveraging that platform built through Claude to profile and target voters in Malaysia, constituency by constituency, using census and electoral data that they had ingested," the report said.

报告称,但该行动被阻止,“没有证据表明其渗透到了真实的社群中”。

But the campaign was stopped "without evidence of breakout into authentic communities", it said.

报告未说明该行动何时开始或持续了多久,但详细介绍了Anthropic在2025年12月至2026年8月期间为打击其人工智能模型被滥用所做出的努力。

The report did not specify when the operation began or how long it lasted, but detailed Anthropic's efforts to disrupt misuse of its AI models between December 2025 and August 2026.

通讯部表示,马来西亚正在起草一项法案,以“加强该国的人工智能治理框架”。

The communications ministry said Malaysia was drafting a bill to "strengthen the country's AI governance framework".

该部表示:“政府将继续加强对滥用人工智能技术风险的监控,并加强与技术提供商、数字平台及相关机构的合作,以应对网络危害,并确保该国的数字环境保持安全、透明和受信。”

"The government will continue to strengthen monitoring of the risks of misuse of AI technology as well as collaboration with technology providers, digital platforms and relevant agencies to address online harm and ensure that the country's digital environment remains safe, transparent and trusted," it said.

阅读媒体原文 ↗