AI短劇《桃花簪》，被全面下架！

近日，有網(wǎng)友發(fā)現(xiàn)自己發(fā)布在網(wǎng)絡(luò)上的個人漢服寫真被一部AI短劇《桃花簪》盜用，抓取復(fù)刻其形象之后生成劇中反派角色，相關(guān)內(nèi)容在多平臺傳播。

網(wǎng)友“白菜漢服妝造”（以下簡稱“白菜”）稱遭遇AI肖像權(quán)侵權(quán)，其發(fā)布在網(wǎng)上的個人漢服寫真被AI短劇《桃花簪》擅自抓取復(fù)刻，生成反派角色，惡意丑化人設(shè)，涉事內(nèi)容在紅果短劇、抖音等平臺傳播，“白菜”已委托律師正式維權(quán)。

網(wǎng)友的個人漢服寫真形象（右）被短劇盜用（左）

As AI-generated short dramas surge across Chinese streaming platforms, ordinary users are discovering their faces lifted from social media and used – without consent, without credit, and sometimes as villains – in productions they've never heard of.

A beauty blogger with only a few dozen followers, who goes by the name Cabbage, said he was stunned to find a character in the AI short drama Peach Blossom Hairpin that looked "almost identical" to him – from his outfit and accessories to his makeup.

The character, he said, was portrayed as a crude, greedy figure with exaggerated features and disturbing behavior.

該網(wǎng)友的維權(quán)信息一經(jīng)發(fā)布，便引起廣泛討論，并于3月31日登上微博熱搜。

4月3日，“紅果短劇官方賬號”微信公眾號發(fā)布聲明回應(yīng)此事稱，已認(rèn)定出品方違反相關(guān)內(nèi)容合規(guī)使用規(guī)定，構(gòu)成違規(guī)違約，即全面下架《桃花簪》，并暫停該出品方上傳所有劇集15天。

聲明表示，平臺鼓勵權(quán)利人和用戶舉報(bào)監(jiān)督。信息一經(jīng)確認(rèn)，平臺將第一時間處理。舉報(bào)郵箱：hongguojubao@bytedance.com。

In response, Hongguo, a popular micro-drama platform, issued a statement on Friday saying its probe found that the producer had violated content rules. The platform removed Peach Blossom Hairpin and suspended the producer from uploading for 15 days.

據(jù)此前報(bào)道，被“盜臉”的網(wǎng)友并非個例。同是在《桃花簪》中，模特“七海”發(fā)現(xiàn)自己的形象也被盜用、塑造成大反派：滿臉麻子、毆打他人、虐待動物。“七海”表示，這極大傷害了她的人格，她已向北京筆墨留香科技有限公司（紅果短劇平臺運(yùn)營主體）致律師函。

上述兩位涉事人的個人形象都在事情被曝光后不久被緊急替換。

模特“七海”的個人形象被AI短劇盜用

中國政法大學(xué)知識產(chǎn)權(quán)中心特約研究員、北京嘉濰律師事務(wù)所合伙人趙占領(lǐng)表示，從法律角度看，無論是否主動使用博主照片，在侵權(quán)定性上差異不大，核心都在于是否未經(jīng)許可使用了具有可識別性的個人形象。根據(jù)民法典及司法實(shí)踐，肖像權(quán)侵權(quán)的認(rèn)定標(biāo)準(zhǔn)是可識別性，而非技術(shù)手段。只要AI生成的角色，能讓公眾聯(lián)想到特定的真實(shí)人物，就可能構(gòu)成侵權(quán)。

趙占領(lǐng)同時表示，假如制作方直接復(fù)制博主照片進(jìn)行AI處理，主觀惡意明顯，屬于典型的以信息技術(shù)手段偽造、制作他人肖像，構(gòu)成肖像侵權(quán)。

Zhao Zhanling, a researcher at China University of Political Science and Law, said that whether a producer actively uses a blogger's photo or not, the legal standard for infringement remains the same: identifiability.

Under China's Civil Code, if an AI-generated character leads the public to associate it with a real person, it may constitute a portrait rights violation. Directly copying a photo for AI processing amounts to a clear case of infringement.

當(dāng)懷疑自己被“盜臉”時怎么辦？

趙占領(lǐng)認(rèn)為，AI時代普通人面對“盜臉”等現(xiàn)象，維權(quán)確實(shí)比以往更加困難。

趙占領(lǐng)建議，首先，做好證據(jù)固定，第一時間對侵權(quán)內(nèi)容進(jìn)行錄屏、截圖，最好通過時間戳、區(qū)塊鏈等方式保全證據(jù)；其次，向平臺進(jìn)行投訴，依據(jù)通知刪除規(guī)則要求平臺下架侵權(quán)內(nèi)容，這往往是成本最低、見效最快的方式；最后，如果情節(jié)嚴(yán)重，可提起法律訴訟，要求侵權(quán)方承擔(dān)法律責(zé)任。

When users suspect their face has been stolen, Zhao advises three steps: First, secure evidence through screenshots or blockchain timestamps; second, file a complaint with the platform under the notice-and-takedown rule — the cheapest and fastest option; and third, pursue legal action if the case is serious.

來源：長安街知事 央視新聞 中國青年報(bào) 新京報(bào) 上海日報(bào)

跟著China Daily

精讀英語新聞

“無痛”學(xué)英語，每天20分鐘就夠！