后患无穷的Deepfake——
我们中隐藏着AI造人...快躲好！
AI造脸和视频伪造
全文字数：1944字 / 阅读时间：6分钟
AI换脸，你认得出么？
斯内普教授换脸沈叔叔，来源哔哩哔哩
那张笑脸可能是计算机生成的假货
At first glance, Ramsey's profile looks like many others on LinkedIn: the bland headshot with a slightly stiff smile; a boilerplate description of RingCentral, the software company where she says she works; and a brief job history.
乍一看，这位“拉姆齐小姐”的个人资料和LinkedIn职场社交平台上的许多人一样：普通的大头照，略微僵硬的笑容；对于自己的公司RingCentral的标准描述；以及简洁的工作经历。
Source: Stanford Internet Observatory
Credit: Connie Hanzhang Jin/NPR
资料来源：斯坦福互联网天文台
图片来源：Connie Hanzhang Jin/NPR
But there were oddities in the photo: the single earring and strange hair, The central positioning of the eyes, the blurry background. Alone, any of these clues might be explained away, but together, they aroused Josh Goldstein's suspicions：it’s a fake face made by AI.
但照片中有一些奇怪之处：只带了一边的耳环，不自然的头发，她处于正中心的双眼的位置，模糊的背景。这些线索中的任何一条单独来看都没什么问题，但它们同时出现就引起了Josh Goldstein的怀疑：这是一张由AI生成的假脸。
Source: Stanford Internet Observatory
Credit: Connie Hanzhang Jin/NPR
资料来源：斯坦福互联网天文台
图片来源：Connie Hanzhang Jin/NPR
完美混入人群
Hany Farid, an expert in digital media forensics at the University of California, Berkeley，found faces made by AI have become "indistinguishable" from real faces. People even consider computer-made faces slightly more trustworthy than real ones. Farid suspects that's because the AI sticks to the most average features when creating a face.
加州大学伯克利分校的数字媒体取证专家汉·法里德发现，AI生成的人脸与真实人脸“难辨真伪”。人们甚至认为电脑制作的脸比真实的脸更可信。法里德猜测，这是因为AI在创建人脸时坚持选用最平凡的外貌特征。
Source: Stanford Internet Observatory
Credit: Connie Hanzhang Jin/NPR
资料来源：斯坦福互联网天文台
图片来源：Connie Hanzhang Jin/NPR
Although some fake accounts have pushed Chinese disinformation; harassed activists; and masqueraded as Americans supporting former President Donald Trump and independent news outlets spreading pro-Kremlin propaganda，NPR found that many of the LinkedIn profiles seem to have a far more mundane purpose: drumming up sales for companies big and small.
虽然有假装成活人的账号发表了有关中国的虚假信息；骚扰了各种活动人士；伪装成了支持前总统特朗普的美国人和宣传亲俄风气的独立新闻机构等等，但美国全国公共广播电台（NPR）发现，LinkedIn上的许多假个人资料似乎有一个更“平凡”的目的：为大大小小的公司推销产品。
amtitus/Digital Vision Vectors/Getty Images
amtitus/数字视觉矢量/盖蒂图像
Demand for online sales leads exploded during the pandemic as it became hard for sales teams to pitch their products in person. By using fake profiles, companies can cast a wide net online without beefing up their own sales staff or hitting LinkedIn's limits on messages.
疫情期间，销售团队很难面对面推销产品，因而在线销售的需求激增。通过使用虚假的个人资料，公司可以在网上广撒网，而不必扩充销售团队；每个账号也不必因为发送太多信息而被LinkedIn限制。
It’s difficult to turn down the chance: from a business perspective, making social media accounts with computer-generated faces has its advantages: It's cheaper than hiring multiple people to create real accounts, and the images are convincing.
AI假脸让公司难以拒绝：从商业角度来看，使用计算机生成的人脸制作社交媒体账户有其优势——比雇佣多人创建真实账户更便宜，而且人像逼真，令人信服。
穷鸟触笼！还有下一步
Computer-generated faces are "the canary in the coal mine," Farid said. "It's the beginning of what is coming next, which is full blown audio-video deepfakes targeted to a specific person."
法里德称，AI生成人脸让我们陷入“穷鸟触笼”的危险局面。“这只是开始，下一步是针对特定目标的全方位音频-视频的深度伪造。”
Source：Facebook
For instance, the heavily manipulated video that circulated on social media recently purporting to show Ukrainian President Volodymyr Zelenskyy calling on his soldiers to surrender.
比如最近在社交媒体上流传的一段深度伪造的视频，该视频声称乌克兰总统泽连斯基呼吁本国士兵投降。
深陷猜疑链？
"The deepfake is not very well done," said Sam Gregory of the human rights group Witness, which specializes in detecting inauthentic media in crises.He said Zelenskyy himself quickly saying it was a fake, helped slow its spread in the West.But he also said that despite the deepfake not being particularly sophisticated, it should still be considered dangerous.
人权组织“目击者”（Witness）的萨姆·格雷戈里表示：“深度伪造并不完美。”该组织专门调查危机中的虚假媒体。他表示，泽连斯基本人很快表示这是一段假的视频，减缓了该视频在西方的传播。但他也称，尽管深度伪造不是特别复杂，却仍然有危险性。
Ben Margot/AP
本·玛戈特/美联社
The video, regardless of where it came from and its quality, could potentially make some people question the veracity of videos of Zelenskyy in the future. "The particular issue is also around the liar's dividend, where it's easy to claim a true video is falsified and place the onus on people to prove it's authentic," Gregory said.
无论视频来自何处、质量如何，未来都可能会有人质疑泽连斯基视频的真实性。格雷戈里说：“这类问题还关乎骗子红利，很容易诱导人们认为真实的视频是伪造的，并让人们要求证明其真实性。”
冰山一角
"The next time the president goes on television, some people might think, 'Wait a minute — is this real?' " Farid added: "It pollutes the information ecosystem, and it casts a shadow on all content，" he said.
法里德补充道：“深度伪造污染了信息生态系统，给所有网络内容蒙上了阴影。”他说，“下次总统上电视时，有些人可能会想：‘等一下，这是真的吗？’”
"Now you have the perfect storm," he said. "I can create this content easily, inexpensively and quickly, I can deliver it en masse to the world, and I have a very willing and eager public that will amplify that for me."
“深度伪造会带来巨大的信息灾难。”他表示，“我可以轻松、廉价、快速地向全世界发布信息，而那些对此饶有兴趣的公众，会为我加大传播的力度。”
杀鸡焉用牛刀
Despite people's fears, sophisticated, deceptive videos known as "deepfakes" haven't arrived this political season. But it's not because they aren't a threat, sources tell NPR. It's because simple deceptions like selective editing or outright lies have worked just fine.
尽管人们很担忧，但在当下的政治季，那些经过深度伪造的复杂、欺骗性的视频还没有粉墨登场。而消息人士告诉NPR，这并不是因为深度伪造已经构不成威胁。而是因为更简单的欺骗（如选择性编辑或直接撒谎）效果也很好。
重点词汇
The canary in the coal mine 穷鸟触笼
直译为“煤矿中的金丝雀”。金丝雀对瓦斯十分敏感，只要矿坑内稍有一丝丝瓦斯，它便会焦躁不安，甚至啼叫，因此以前矿工们都会在矿坑里放金丝雀，当作早期示警的工具。现比喻危险的局面。
汉语中有一成语为“穷鸟触笼”有类似的意思，语出《旧唐书》：“既知四隅断绝，百计奔冲，如穷鸟触笼，似飞蛾赴焰。”
原文链接：
https://www.npr.org/2018/02/03/582767531/in-an-era-of-fake-news-advancing-face-swap-apps-blur-more-lines
https://www.npr.org/2022/03/27/1088140809/fake-linkedin-profiles
https://www.npr.org/2022/03/16/1087062648/deepfake-video-zelenskyy-experts-war-manipulation-ukraine-russia
https://www.npr.org/2020/10/01/918223033/where-are-the-deepfakes-in-this-presidential-election
编译 | 陈宇嘉
排版 | 陈宇嘉
