Global EditionASIA 中文双语Français
Latest News

AI deepfakes target Chinese celebrities, spark public outcry

By Liu Hui | chinadaily.com.cn | Updated: 2025-05-21 17:13
Share
Share - WeChat

5月18日,演员邓超与孙俪通过微博发布联合声明,指出网络平台出现多支未经授权的AI伪造代言视频。

Chinese actors Deng Chao and his actress wife Sun Li issued a joint statement on social networking platform Sina Weibo Sunday, denouncing a series of unauthorized AI-generated videos circulating online that falsely feature those endorsing products.

这些视频通过技术手段合成二人的肖像和声音,假冒采访形式进行虚假商业宣传。

The videos used advanced AI to replicate their faces and voices, inserting them into fabricated interview formats to promote products without their permission.

A screenshot from Sina Weibo shows actress Sun Li denouncing multiple AI-generated fake endorsement videos with a strong warning: "Don't believe it! It's all fake".

孙俪在转发声明时连用三个"不要相信"和三个"都是骗人的"严正表态,相关话题迅速登上热搜。

Sun's emphatic reposting of "Don't believe it" and "It's all fake" three times quickly made the issue a top trending topic on Weibo.

技术专家指出,现有AI语音克隆技术仅需采集少量声音样本即可在数秒内完成高度逼真的声纹复制,其操作门槛已降至类似手机修图软件的简易程度。

Some tech experts warn that AI voice cloning is dangerously easy, requiring only a few seconds of audio to create highly realistic imitations. The process is as simple as using a mobile photo-editing app, lowering the barrier for abuse.

A Douyin screenshot shows a deepfake video falsely using Quan Hongchan's voice to endorse rural honey products.

今年以来,这一技术被大规模滥用,已有多位公众人物遭遇AI造假。4月,奥运跳水冠军全红婵父亲发文称,发现多个账号利用AI技术篡改女儿东京奥运会采访视频,将其原声替换为蜂蜜产品推销内容。去年12月,伪造的"张文宏推荐蛋白棒"视频因高度逼真导致大量消费者受骗。

The misuse of AI voice cloning has surged this year. This April, the father of China's diving phenom Quan Hongchan reported that his daughter's Tokyo Olympics interview audio was deepfaked to push honey ads. Similarly, last December, a convincing deepfake video of epidemiologist Zhang Wenhong endorsing protein bars also led to consumer scams.

在抖音、快手等短视频平台,此类AI造假已形成产业链。二手交易平台提供低至10元的“声音克隆”服务,包含多位明星声线模型。

A booming AI deepfake industry is thriving on short-video sharing platforms such as Douyin and Kuaishou, with voice cloning starting at just 10 yuan ($1.38) on secondhand markets, offering celebrity voices.

业内人士透露,专业团队采用标准化流程:采集公开影像、AI声纹克隆、视频合成,日均产能可达上百条。这些视频针对母婴、保健品等领域精准投放,在抖音、快手、小红书等平台持续传播。

Industry experts reveal that professional teams can produce hundreds of these synthesized videos daily by collecting public footage, cloning voiceprints, and then generating the content. These deepfakes are often used to target specific markets such as health supplements and baby products, spreading rapidly across Douyin, Kuaishou, and Xiaohongshu (RedNote).

据《咸宁日报》报道,AI拟声带货的商品存在资质不明、厂商不清等问题,涉事店铺多为2024年新注册账号,且平台未对AI生成内容进行有效标识。

According to Xianning Daily, many products promoted by AI voices lack verified credentials, often sold by accounts newly registered in 2024. A major issue is that platforms are failing to clearly label this AI-generated content.

在今年全国两会期间,全国人大代表、小米集团创始人雷军提交了关于规范"AI换脸拟声"技术的建议。

During this year's two sessions, NPC deputy Lei Jun, also Xiaomi's founder, suggested regulating "AI face-swapping and voice-cloning" technologies.

雷军表示,自己就曾遭遇AI技术滥用——网络上流传着大量使用其声音和形象的伪造视频,这些"假雷军"用标志性的湖北口音吐槽堵车、调侃调休,逼真程度令许多网友难辨真假,甚至在其社交账号下留言求证"是否本人"。

Lei revealed he's also a victim, with many fake videos featuring his voice and likeness spreading online. These clips feature a "fake" Lei Jun speaking in his distinctive Hubei accent, humorously discussing traffic jams and holiday schedule changes. The videos are so realistic that many viewers couldn't tell the difference, even asking Lei on social media, "Is this really you?"

目前,国内正在立法上重视此类问题。2024年4月,北京互联网法院宣判全国首例“AI声音侵权案”,首次明确自然人声音权益的保护范围可及于AI生成内容。

China is stepping up its fight against AI misuse. Last April, the Beijing Internet Court ruled in the country's first AI voice infringement case, extending legal protection to natural voices in AI-generated content.

此外,类似“白名单制度”也会在9月1日上线,让以往难以辨别的AI作品“持证上岗”。

A new "whitelist system" for AI content, requiring proper labeling, is set to launch on September 1.

对平台而言,当务之急是提升技术手段,对AI合成内容进行识别和标注,“用AI打败AI”,及时清理虚假信息,对违规账号进行封禁处理。

Platforms urgently need to detect and label AI-generated media—"fighting AI with AI"—and promptly remove fakes while banning offenders.

作为普通消费者,在AI合成日益被滥用的趋势下,尤其要提高辨别能力和防范意识,不要轻信、传播未经证实的消息,更不要轻易地“一键下单”。

Consumers, in turn, should remain cautious as AI abuse spreads. Don't trust or share unverified content, and think twice before impulse buying from suspicious "one-click" offers.

Today's Top News

Editor's picks

Most Viewed

Top
BACK TO THE TOP
English
Copyright 1995 - . All rights reserved. The content (including but not limited to text, photo, multimedia information, etc) published in this site belongs to China Daily Information Co (CDIC). Without written authorization from CDIC, such content shall not be republished or used in any form. Note: Browsers with 1024*768 or higher resolution are suggested for this site.
License for publishing multimedia online 0108263

Registration Number: 130349
FOLLOW US