算力商业
futurism.com ·
@Victor Tangermann
Kalshi 承认用 AI 将他人视频改造成广告并替换其族裔
科技 YouTuber Elliott Choy 发现自己的旧视频被 Kalshi 用 AI 深度伪造技术改造成广告，视频中的亚裔创作者被替换成不存在的白人形象。NPR 记者 Bobby Allyn 向 Kalshi 求证后，公司代表承认该广告真实存在，但称其“已过时”且不再使用 AI 生成广告，并正在审查制作该广告的代理机构。事件与斯坦福大学此前用 AI 替换照片中学生的案例相继曝光，凸显 AI 深度伪造在商业营销中的滥用及内容审核漏洞。
AI推荐理由 报道了 Kalshi 被指用 AI 深度伪造技术将 YouTuber Elliott Choy 的视频改造成广告并改变其族裔的具体事件，NPR 记者 Bobby Allyn 已向公司求证，公司承认广告真实但称已停用 AI 生成广告并审查合作机构。这是 AI 深度伪造在商业广告中滥用的可验证案例，涉及内容版权、族裔替换和平台审核等实际风险，但事件本身属伦理争议而非技术突破，且仍有涉事机构、受害规模等关键信息缺失。
62
正文
Elliott Choy via YouTube
As the public backlash to cheap AI-generated slop continues to grow, companies continue to insist on cutting costs by relying on the tech for their ads.
Some are taking the trend to the extreme by creating deepfakes of videos uploaded by real influencers to market their products.
Sleazy gambling and insider trading platform Kalshi is under fire after tech YouTuber Elliott Choy found that one of his videos had been turned into an ad that he was served on YouTube — a process that involved turning the Asian-American creator “into a white dude” that doesn’t exist.
It’s a particularly egregious example that not only involves a company shamelessly stealing user-generated content, but whitewashing the subject in the video as well, all the while making them shill a product. As Choy notes, he made the discovery just hours after Stanford University was caught using AI to swap out a student in a photo and replace him with a non-existent Black woman.
Ironically, Choy has been sharing ads for AI company Anthropic’s Claude assistant.
The original video Kalshi ripped off was about Choy moving into a new apartment in New York City and was originally uploaded in September of last year.
NPR correspondent Bobby Allyn has since confirmed that the offending Kalshi ad is very much real. A representative for the company told him that it’s “old and that they no longer use AI-generated ads.” The rep also said that it’s “‘reviewing our relationship with’ the agency that made it.”
Plenty of questions remain. We still don’t know which agency Kalshi contracted for the ad or how many influencers had their content ripped off by the company — or if the company intends to proactively make things right with other victims.
Allyn first noticed the trend in July of last year. Videos on TikTok clone existing user-generated content using deepfake tech, racking up tens of millions of views. However, most of these videos were seemingly created by scammers — not by a mainstream company looking to generate ads.
“It’s the kind of thing that would be super-easy to do with today’s AI tools and something that would easily slip through the content moderation cracks,” University of California digital forensics expert Hany Farid told Allwyn at the time.
Kalshi has been experimenting with generative AI for its marketing for quite some time. Last year, it aired an AI slop ad for the NBA Finals that the company claimed cost just $2,000 to make — though it’s debatable whether the company got its money’s worth, judging by the result.
Since then, as the backlash to AI slop has grown astronomically, the company has been actively trying to play the other side, hiring A-listers including the actor Timothée Chalamet and Oscar-winning cinematographers.
“I saw pretty much every brand under the sun use AI,” Kalshi’s head of growth Keaton Inglis told Ad Age earlier this month. “And I transitioned us, going from being a leader in AI to as human… as possible going forward.”
Considering the company’s offensive, race-swapping AI ads are being pushed on unsuspecting users on social media, that promise feels mightily disingenuous.
More on Kalshi: Kalshi Just Issued Its First-Ever Lifetime Ban to the Funniest Person You Can Possibly Imagine