AI正在制作扭曲现实的中期选举广告，内容涉及被涂鸦的圣经、高档晚餐以及一张印有“Death to America”字样的桌游卡牌。
AI is producing reality-bending midterm ads about defaced Bibles, upscale dinners and a board game with a "Death to America" playing card.
为何重要：关于深度伪造的警告正在成为现实。选举法律未能跟上形势——在这个世界里，竞选活动可以让对手说出他们从未说过的话，并制造出从未发生的逼真场景。
Why it matters: The warnings about deepfakes are coming true. Election laws haven't kept pace with a world where campaigns can put words in opponents' mouths and create realistic-looking scenes that didn't happen.
卫斯理媒体项目（Wesleyan Media Project）发现，截至10月11日，已有1.37亿美元用于制作220条AI生成的政治广告，其中大部分来自共和党。
The Wesleyan Media Project found $137 million had been spent on 220 AI-generated ads through Oct. 11, mostly from Republicans.
- 四个州有法律禁止在政治广告中使用某些AI生成的深度伪造内容，而另外27个州要求某种形式的披露其使用情况。
• Four states have laws prohibiting certain uses of AI-generated deepfakes in political ads, while 27 more require some form of disclosure about their use.
- 没有联邦法律专门规范AI生成的政治广告。
• No federal law specifically regulates AI-generated political ads.
大局观： Axios的分析发现，众多政治广告部分或全部由生成式AI制作，有时未予披露。
The big picture: An Axios analysis found numerous political ads partly or fully made by generative AI, sometimes undisclosed.
佐治亚州参议院：参议院领导基金（Senate Leadership Fund）——一个与参议院共和党人结盟的超级政治行动委员会——已推出一系列AI生成的深度伪造广告，描绘民主党参议员Jon Ossoff接受美容护理。
Georgia Senate: Senate Leadership Fund, a super PAC aligned with Senate Republicans, has run a series of AI-generated deepfake ads depicting Democratic Sen. Jon Ossoff getting beauty treatments.
- 这些广告嘲讽Ossoff据称在竞选中用于发型和化妆的开支，称他“说话好听”但“有丑陋的记录”和“需要掩盖很多东西”。
• The ads mock Ossoff's reported campaign spending on hair and makeup, saying he "talks pretty" but has "an ugly record" and "a lot to cover up."
- Ossoff在C-SPAN采访中表示：“看到……深度伪造视频被用来编造政治广告，令人深感不安。这正在破坏政治辩论中的真相，该广告应立即下架。”
• Ossoff said in a C-SPAN interview: "It is deeply troubling to see ... deepfake videos used to fabricate political advertising. And it's a s toward the destruction of truth in political debate, and the ad should come down immediately."
- 这些广告未披露使用AI，而佐治亚州法律并未要求如此。
• The ads don't disclose AI use, which Georgia law doesn't require.
密歇根州参议院：一个名为Front Line Action的超级政治行动委员会在一则广告中使用AI，描绘民主党候选人Abdul El-Sayed与播客主持人Hasan Piker以及四名戴着尼卡布的女性合影的虚构家庭照。
Michigan Senate: A super PAC called Front Line Action used AI in an ad depicting Democratic candidate Abdul El-Sayed in a fabricated family photo with podcaster Hasan Piker and four women wearing niqabs.
- 另一个AI生成的场景显示，El-Sayed与他的同父异母姐姐Eman Abdelhadi以及Piker一起玩一种类似《风险》（Risk）的桌游。该场景中的卡片上印有包括“Death to America”、“不要庆祝感恩节”和“庆祝9/11”等陈述。
• Another AI-generated scene has El-Sayed playing a Risk-style board game with his half-sister, Eman Abdelhadi, and Piker. It depicts cards bearing statements that include "Death to America," "Do not Celebrate Thanksgiving" and "Celebrate 9/11."
- 该团队为该广告进行了辩护，对CNN表示该广告“基于赛义德本人的公开声明以及有关他及其竞选活动相关或支持者的有据可查的报道和记录”。
• The group has defended the ad, telling CNN it is "grounded in El-Sayed's own public statements and documented reporting and records concerning him and individuals associated with or supportive of his campaign."
- 该广告既有书面披露，也有口头披露，称其“完全或部分使用人工智能制作”。
• The ad features both a written and spoken disclosure that it was "created in whole or in part with the use of artificial intelligence."
纽约州长：共和党人布鲁斯·布拉克曼的竞选团队及其关联的超级政治行动委员会反复在攻击民主党州长凯西·霍赫尔的广告中使用生成式人工智能，披露程度各不相同。
New York governor: Republican Bruce Blakeman's campaign and allied super PACs have repeatedly used generative AI in ads attacking Democratic Gov. Kathy Hochul, with varying degrees of disclosure.
- 布拉克曼竞选团队发布的一则人工智能生成的广告描绘了牙仙在努力筹集现金给孩子们，她说她“就在凯西·霍赫尔的纽约州，负担不起当牙仙的费用”。
• One AI-generated ad released by the Blakeman campaign depicts the Tooth Fairy struggling to come up with cash to give to children, saying she "just can't afford to be the Tooth Fairy in Kathy Hochul's New York."
- 另一则布拉克曼的广告画面上显示：“我们问人工智能，一则真实的凯西·霍赫尔广告会是什么样子。”该广告包含人工智能生成的场景：霍赫尔参加一场高档晚宴，并在警察局外站立，而一名男子问她：“要不要来点可卡因？”
• Another Blakesman spot says onscreen: "We asked AI what a truthful Kathy Hochul ad would look like." The ad features AI-generated scenes of Hochul attending an upscale dinner and standing outside a police precinct while a man asks her: "Want some crack?"
宾夕法尼亚州长：共和党候选人斯泰西·加里蒂的新广告中出现了民主党州长乔什·夏皮罗的人工智能生成版本，他在广告中谈论税收、堕胎和数据中心。
Pennsylvania governor: new ad from Republican candidate Stacy Garrity features an AI-generated version of Democratic Gov. Josh Shapiro talking about taxes, abortion and data centers.
- 广告让他说出这句话：“这则广告是用人工智能制作的，多亏了那些宏大而美丽的数据中心。事实上，我就是那个在宾夕法尼亚州这里加速推进数据中心建设的人，就在他们递给我500万美元政治捐款之后。”
• The spot puts this quote in his mouth: "This ad was created with AI, thanks to big, beautiful data centers. Fact is, I was the one who fast-tracked data centers right here in PA, right after they handed me $5 million in political contributions."
- 夏皮罗的发言人曼努埃尔·邦德在向Axios的声明中驳斥该广告是“廉价、不诚实的人工智能废话”。
• Shapiro spokesperson Manuel Bonder, in a statement to Axios, dismissed the ad as "cheap, dishonest AI bullshit."
民主党候选人詹姆斯·塔拉里科一直是共和党肯·帕克斯顿竞选团队及其外部超级政治行动委员会制作的人工智能生成广告的常见目标。
Democratic candidate James Talarico has been a consistent target of AI-generated ads from both Republican Ken Paxton's campaign and outside super PACs.
- 帕克斯顿竞选团队的一则广告使用人工智能描绘塔拉里科用蜡笔在圣经上涂鸦，并说：“起初，上帝是非二元的，他也不是基督徒。上帝说，‘愿有六种性别’，于是就有了六种性别。”
• A Paxton campaign ad uses AI to depict Talarico defacing a Bible with crayon and saying, "In the beginning God was non-binary, and he wasn't a Christian. And God said, 'let there be six genders,' and there were."
- 塔拉里科告诉NBC，他“曾在立法机构试图禁止在政治竞选中使用AI深度伪造视频，因为它们具有误导性。它们在歪曲本次竞选中的候选人。”
• Talarico told NBC he "tried in the legislature to ban AI deepfake videos in political campaigns because they are misleading. They are lying about a candidate in this race."
肯塔基州第6选区：由特朗普总统支持的MAGA Inc.支持的No Going Back PAC已经投放了多则利用AI提升共和党国会候选人拉尔夫·阿尔瓦拉多的广告。
Kentucky's 6th District: No Going Back PAC, backed by President Trump's MAGA Inc., has run several ads using AI to boost Republican congressional candidate Ralph Alvarado.
- 一则广告包含多个AI生成的深度伪造视频，其中一幕显示民主党人扎克·丹博凝视着镜头。
• One ad features multiple AI-generated deepfakes of Democrat Zach Dembo, including one scene in which he stares into the camera.
- “我原本以为他们会歪曲我的履历，但我没想到他们会歪曲我的外貌，”丹博在向Axios的声明中表示，并补充说：“利用AI制造虚假或误导性图像是我们民主走向危险道路的开始。”
• "I expected them to lie about my record, but I did not expect them to lie about my appearance," Dembo said in a statement to Axios, adding that "using AI to create fake or misleading images is the beginning of a very dangerous path for our democracy."