字号·· | 护眼
每日电讯报

AI垃圾宣传正在为中期选举加速The AI slopaganda turbocharging the midterms

点「原文对照」整页切到原文,或双击某段只看那段的原文。

今年夏天,你可能看过马特·达蒙在克里斯托弗·诺兰近三小时的史诗电影中饰演奥德修斯。但你看过由密歇根州民主党参议员候选人主演的那个版本吗?

You may have seen Matt Damon as Odysseus in Christopher Nolan’s near-three-hour epic this summer. But what about the one starring Michigan’s Democratic nominee for Senate?

在这部对著名故事的离奇演绎中,一匹特洛伊木马被拖过一座桥,上面挂着一条横幅,写着:“绝对不是共产主义。”

In this bizarre interpretation of the famous tale, a Trojan horse is dragged across a bridge, adorned with a banner which reads: “Definitely not communism.”

木马内部,密歇根州的阿卜杜勒·埃尔-赛义德身穿全套特洛伊军装,诡秘地笑着对同为民主社会主义者的佐赫兰·马姆达尼、伯尼·桑德斯和亚历山德里娅·奥卡西奥-科尔特斯说:“他们永远看不到我们来了。他们真以为我们只是民主党人。”这段视频由埃尔-赛义德的对手迈克·罗杰斯于8月发布,罗杰斯是获得特朗普支持的共和党候选人,正在角逐密歇根州竞争激烈的参议员席位。他的竞选团队警告称,这位民主党人持有所谓的“极端主义立场”。

Inside, Michigan’s Abdul El-Sayed is dressed in full Trojan military garb and, laughing conspiratorially, tells his fellow democratic socialists Zohran Mamdani, Bernie Sanders and Alexandria Ocasio-Cortez: “They’ll never see us coming. They really think we’re just Democrats.”The video was posted in August by Mr El-Sayed’s opponent, Mike Rogers, the Trump-backed Republican candidate for Michigan’s hotly contested Senate seat. His campaign has warned of the Democrat’s supposed “extremist positions”.

“他把我塑造成了奥德修斯,”埃尔-赛义德在被问及这则广告时说。“简直就是《奥德赛》里的英雄。我接受。”这只是所谓“AI垃圾”政治广告泛滥的其中一个例子,在周二中期选举之前,这类广告已充斥美国各地选民的屏幕。此类广告的支出急剧膨胀,即将到来的选举标志着生成式AI技术首次在美国重大选举中无处不在。

“He made me Odysseus,” Mr El-Sayed said when asked about the ad. “Like, literally the hero of The Odyssey. I’ll take it.”It is just one example of a slew of so-called “AI slop” political ads that have flooded the screens of voters across the US ahead of the midterm elections on Tuesday. Spending on these ads has ballooned, with the coming polls marking the first major US elections in which generative AI technology has been ubiquitous.

追踪选举广告支出的卫斯理媒体项目(WMP)本月早些时候发布的一份报告发现,仅今年一年,就有1.37亿美元(1.03亿英镑)用于AI生成的广告,而2024年总统选举年这一数字仅为5万美元。

A report published earlier this month by the Wesleyan Media Project (WMP), which tracks election ad spending, found that $137m (£103m) has been spent on AI-generated adverts this year alone, up from just $50,000 during the 2024 presidential election year.

另一则针对埃尔-赛义德的AI广告由一个与得克萨斯州共和党参议员特德·克鲁兹结盟的团体出资,模仿了一部名为《家庭价值观》的情景喜剧。

Another AI ad targeting Dr El-Sayed, paid for by a group aligned with Ted Cruz, a Texan Republican senator, depicted a mock sitcom called Family Values.

“摧毁以色列和西方对他们来说不仅仅是一场游戏——而是一项使命,”一名旁白在欢快的音乐声中说道,画面蒙太奇中穿插着埃尔-赛义德与家人以及颇具争议的左翼主播哈桑·皮克一起大笑和玩桌游的片段,皮克曾与这位民主党人一起竞选。

“The destruction of Israel and the West isn’t just a game for them – it’s a mission,” a narrator says over jaunty music and a montage of Dr El-Sayed laughing and playing board games with family members and the controversial Left-wing streamer Hasan Piker, who campaigned alongside the Democrat.

这则广告试图将埃尔-赛义德博士的家族与伊斯兰组织穆斯林兄弟会联系在一起,随着竞选在最近几周进入白热化阶段,这一说法成为了一个引爆点。唐纳德·特朗普上周称他是“圣战分子”,“站在恐怖分子一边”。埃尔-赛义德博士已多次否认其家族与穆斯林兄弟会有任何关联。

The advert attempts to tie Dr El-Sayed’s family to the Islamist group the Muslim Brotherhood, a claim which became a flashpoint as the campaign reached fever pitch in recent weeks. Donald Trump last week called him a “jihadist” who was “on the side of the terrorists”. Dr El-Sayed has repeatedly denied any family ties to the Muslim Brotherhood.

“埃尔-赛义德不是在玩游戏,他的家庭价值观不适合密歇根,”旁白说道,广告结尾是一张由AI生成的图像,显示这位民主党候选人与戴着头巾的我们的女性以及皮克先生合影,皮克先生曾因说美国“活该遭遇9·11”而道歉。

“El-Sayed isn’t playing games, and his family values are wrong for Michigan,” the narrator says as the advert ends with an AI-generated image showing the Democrat candidate posing alongside our women in niqabs and Mr Piker, who has apologised for saying America “deserved 9/11”.

埃尔-赛义德并非唯一被针对的人。宾夕法尼亚州民主党州长乔什·夏皮罗周一予以抨击,此前他的对手分享了一段AI生成的视频片段,显示他在任期内面带笑容地列举不利于自己的新闻标题。

Dr El-Sayed isn’t the only one who has been targeted. Josh Shapiro, the Democratic governor of Pennsylvania, hit out on Monday after his rival shared an AI-generated clip showing him grinning while listing unflattering headlines during his incumbency.

这则广告是为共和党人斯泰西·加里蒂的州长竞选制作的,其中出现了数字合成的夏皮罗先生,他对着镜头声称自己“给了你们美国最高的税收之一”,并支持“直到婴儿第一口呼吸前的堕胎权”。

The advert, made for Republican Stacy Garrity’s campaign for governor, featured a digitally created Mr Shapiro claiming to the camera that he had “given you some of the highest taxes in America” and supported “abortion right up to a baby’s first breath”.

夏皮罗先生的一位发言人告诉《每日电讯报》,这则广告是“廉价、不诚实的AI垃圾”,由特朗普为宾夕法尼亚州州长人选推举的人在“她灾难性、苦苦挣扎的竞选最后日子里”推出。

A spokesman for Mr Shapiro told The Telegraph the advert was “cheap, dishonest AI bull----” pushed out by Mr Trump’s pick for Pennsylvania governor in the “closing days of her disastrous, flailing campaign”.

在其他地方,得克萨斯州民主党参议员候选人詹姆斯·塔拉里科与该州总检察长肯·帕克斯顿陷入了一场异常胶着的选战。得克萨斯州自20世纪80年代以来从未将民主党人选入美国参议院,但民调目前显示两位候选人势均力敌。

Elsewhere, James Talarico, the Democratic candidate in Texas for the Senate, is locked in a remarkably tight race against Ken Paxton, the state’s attorney general. Texas has not elected a Democrat to the US Senate since the 1980s, yet polls currently place the two candidates neck-and-neck.

过去几个月里,塔拉里科先生一直是一系列AI攻击广告的目标,这些广告主要集中在他过去关于性别问题的言论上。

Mr Talarico has been the target of a series of AI attack adverts over the past few months, which have focused largely on his past comments on gender issues.

9月,肯·帕克斯顿竞选参议员团队发布了一则广告,画面中由AI生成的塔拉里科正用蜡笔改写《圣经》。他的批注包括“起初,上帝是非二元性别的”以及“上帝说要有六种性别,于是就有了六种性别”等内容。

In September, the Ken Paxton for Senate campaign shared an advert showing an AI-generated Mr Talarico rewriting the Bible with a crayon. His annotations included lines such as “in the beginning, God was non-binary”, and “God said let there be six genders, and there were”.

对此,塔拉里科推出了自己的广告,指责帕克斯顿“腐败”,广告结尾配文称:“由人类制作,非AI生成”。

In response, Mr Talarico put out an advert of his own accusing Mr Paxton of “corruption”, which closed with text stating: “Made by humans, not AI”.

另一则由“理智公民”组织(一个支持特朗普的竞选团体)分享的AI生成广告,则呈现了身着《音乐之声》中玛丽亚·冯·特拉普服装的塔拉里科。

Another AI-generated advert shared by Citizens for Sanity, a pro-Trump campaign group, featured Mr Talarico dressed as Maria Von Trapp from The Sound of Music.

尽管有人呼吁对政治广告中使用AI实施更严格的监管,但美国在很大程度上回避了可能限制言论自由的法规。

Although some have called for stricter regulations on the use of AI in political ads, the US has largely shied away from regulations that might limit freedom of speech.

在第二任期内,美国总统多次分享有争议的AI生成内容,例如一段将奥巴马夫妇描绘成猿猴的种族主义视频,以及本月初发布的另一段视频,画面中他正在追逐并嬉戏那只在约克郡野生动物园咬死一名误入围栏的英国男子的老虎。

During his second term, the US president has repeatedly shared controversial AI-generated content, such as a racist video which depicted the Obamas as apes and another earlier this month in which he chased and played with the tiger which killed a British man who entered an enclosure at Yorkshire Wildlife Park.

批评人士指出,部分广告已偏离政治竞选范畴,进入“深度伪造”领域,即利用AI操纵真实音视频,使人物看起来像说过或做过从未发生的事情。

Critics have claimed that some of the adverts have veered away from political campaigning and into deepfake territory, in which AI is used to manipulate real audio or video, making people appear to have said or done things they never actually did.

威斯康星州国会选区民主党候选人丽贝卡·库克在9月向对手发出停止侵权函,声称对方发布的一段视频属于深度伪造。

Rebecca Cooke, the Democratic candidate for a congressional race in Wisconsin, sent a cease and desist letter to her opponent in September claiming that a video he posted was a deepfake.

威斯康星州共和党众议员德里克·范·奥登使用Grok Imagine制作视频,让库克“说”出想要“尽可能利用腐败,给每个非法移民提供免费医疗”。

Derrick Van Orden, a Republican representative in Wisconsin, used Grok Imagine to make Ms Cooke say that she wanted to “take advantage of as much corruption as possible and give every illegal alien free healthcare”.

威斯康星州共和党国会候选人德里克·范奥登被指控使用人工智能制作其对手的深度伪造视频。她的律师声称,范奥登利用人工智能“制作深度伪造图像,为这些虚假内容披上真实的外衣,目的是伤害库克女士及其声誉”。

Derrick Van Orden, a Republican congressional candidate in Wisconsin, was accused of using AI to make a deepfake video of his opponent Her lawyers claimed that Mr Van Orden had used AI for “deepfake imagery to give the veneer of truth to these falsities, for the purpose of harming Ms Cooke and her reputation”.

范奥登对她的律师函的回应是分享了一段人工智能生成的视频,视频中一只鸭子穿着美国民主社会主义者的T恤说“停止并终止!”。

Mr Van Orden responded to her legal letter by sharing an AI-generated video of a duck saying “cease and desist!” while wearing a Democratic Socialists of America T-shirt.

包括美国政治顾问协会在内的竞选团体呼吁联邦出台法规,对“欺骗性”人工智能内容进行标注,以取代目前各州杂乱不一的限制措施。

Campaign groups including the American Association of Political Consultants are calling for federal regulations labelling “deceptive” AI content to override the current jumble of state restrictions.

迄今为止,大约30个美国州已通过立法,限制使用人工智能生成政治广告,或强制要求披露此类内容。

Roughly 30 US states have so far passed legislation restricting the use of AI to generate political adverts, or forcing its disclosure.

与此同时,有三个州对针对政治候选人、意图影响选举的深度伪造内容进行了限制。马里兰州全面禁止与选举相关的深度伪造内容,而得克萨斯州和明尼苏达州则分别在投票前30天和90天内禁止此类内容。

Meanwhile, three states have limited deepfakes targeting political candidates intended to influence elections. Maryland has banned election-related deepfakes altogether, while Texas and Minnesota have outlawed them within 30 and 90 days of voting respectively.

言论自由倡导者认为,人工智能只是为政治讽刺创造了一种新工具——类似于传统的漫画和讽刺画。

Free speech advocates argue that AI simply creates a new tool for political satire – akin to traditional cartoons and caricatures.

许多针对中期选举候选人的广告聚焦于文化战争议题,并包含笑话和欢快的歌曲,有些人可能会认为这属于讽刺。

Many of the adverts targeting midterm candidates have focused on culture war issues, and included jokes and upbeat songs that some could argue is satire.

一首人工智能乡村歌曲在缅因州电台播出,称该州民主党候选人特洛伊·杰克逊为“怪人”和“令人毛骨悚然的人”,因为“他和自己的表亲生了两个孩子”。

An AI country song aired on Maine radio calling Troy Jackson, the Democratic candidate for this state, a “freak” and a “creep” because “he has two kids with his cousin”.

这首广告歌的歌词中包括:“如果他能操自己的家人,他也能操你的家人。”缅因州共和党参议员苏珊·柯林斯与一则由参议院领导力基金资助的广告划清界限,并呼吁将其撤下。该基金是一个与参议院共和党人结盟的筹款委员会。

Among the lyrics in the jingle: are “If he’ll f--- his own family, he’ll f--- yours too.”Susan Collins, Maine’s Republican senator, distanced herself from one advert funded by the Senate Leadership Fund, a fundraising committee aligned with Senate Republicans, and called for it to be removed.

缅因州共和党参议员苏珊·柯林斯呼吁本党资金方撤下一则广告。人们担忧的不仅是政治对手可能利用人工智能互相攻击,还有海外势力如何运用这项技术散布虚假信息。

Susan Collins, Maine’s Republican senator, called on her own party funders to remove an advert Concerns have not only been raised about how political rivals might use AI to attack each other but also how the technology is employed by overseas powers to sow disinformation.

据《华盛顿邮报》报道,美国国家情报办公室警告称,外国对手正利用人工智能在中期选举前散布虚假信息,俄罗斯在试图破坏选举方面最为活跃。

The US national intelligence office warned that foreign adversaries were using AI to spread disinformation ahead of the midterms, and Russia was the most active in attempts to undermine the election, The Washington Post reported.

研究人员将这些帖子归因于“马特留什卡行动”,该名称源自俄罗斯套娃。这一由俄罗斯国家支持的行动用大量看似来自主流媒体的视频片段淹没社交媒体,以散布虚假信息。

Researchers have attributed the posts to Operation Matryoshka, named after Russian nesting dolls. The Russian state-backed operation swamps social media with clips that appear to be from established media outlets to spread disinformation.

随着技术变得更易获取、更易使用,美国中期选举已成为测试选民是否会被人工智能生成的广告所打动的试验场。

As the technology has become more accessible and easier to use, the US midterms have become a testing ground for whether voters will be moved by AI-generated ads.

韦斯利传媒项目联合主任迈克尔·弗朗茨表示,传统的政治竞选广告“枯燥至极”。

Michael Franz, co-director of the WMP, said that traditional political campaign adverts were as “boring as they come”.

他补充道:“如果你能用一些讽刺、一些滑稽元素来强化那条信息,你就在创造一些更犀利的东西,这可能会让政治广告更有效。”

“If you can enhance that message with some satire, with some goof, you’re creating something sharper which could make political ads more effective,” he added.

阅读媒体原文 ↗