Kalshi is drawing major heat online after creator Elliot Choy accused the company of ripping off one of his videos for an ad — and making his face look like a white man’s.
Choy posted side-by-side stills that show the clear similarities between the appearance of the ad and his video, which depicted him moving into his New York apartment, with his computer and books in the background.
A Kalshi spokesperson said in a statement that the ad was about four months old and made by an agency that supplied the video. The spokesperson said it no longer uses AI-generated content that depicts people and that it’s reviewing its relationship with the agency. They declined to identify the agency.
The episode comes as Madison Avenue — along with other industries — wrestles with how to balance AI’s cost-saving benefits against the risk of pushback from consumers and creative partners.
Kalshi itself is no stranger to controversy. The prediction giant has grown into a multibillion-dollar business by letting retail traders bet on everything from election outcomes to reality TV winners. Kalshi’s disruptive approach has led to numerous lawsuits against the company across the US. State regulators, including New York’s, have argued that Kalshi is running an unlicensed gambling operation, which Kalshi denies.
Choy wasn’t the first creator to call out Kalshi. Another creator, Pushpek Sidhu, posted in June about having a similar experience.
Sidhu, a 29-year-old content creator in Toronto who often posts about food and sports, told Business Insider that someone who identified himself as a Kalshi rep reached out to propose a collaboration. Sidhu said he didn’t hear from the rep again.
Then, while in Houston for a World Cup game, Sidhu said he saw an ad that showed a young man replicating almost completely a video of his own, weighing in on a game, down to the near-identical script and gestures.
As with the Choy video, Sidhu said he was replaced by a white man.
“They could easily pay me a little bit,” Sidhu said of Kalshi.
AI is becoming commonplace in ads
Kalshi was an early mover in using AI for its ads, releasing a spot during the NBA Finals in 2025 that generated industry buzz.
The practice of using AI for social media ads has become commonplace, though it still accounts for a relatively small share of the market.
Resemble AI, a company that tracks deepfakes for government agencies and enterprises, said its global search of Meta’s ads library across a range of industries showed that, on average, about 9% of ads in recent months contain generative AI.
“We’re seeing a ton of ads that are gen AI,” said Zohaib Ahmed, CEO of Resemble AI. “It is being used in almost every industry.”
Apps like Meta and TikTok have policies for labeling AI-generated content.
Companies can employ a range of open-source platforms that are easy to use and can cheaply generate video from images using prompts. Using video-to-video models, companies can create new videos from existing ones, swapping in a different face to reach their target audience.
Ahmed said in the case of the Kalshi ad that Choy flagged, it could have started with a thumbnail from a social video and then swapped in someone else’s face.
He said AI can be used creatively, but the demand for hyper-personalized ads can lead companies to cross ethical lines.
“It’s difficult to draw the line of where you’ve crossed legally,” Ahmed said. “However, there’s a pretty solid ethical line. You are clearly taking someone else’s content and then repurposing it for yourself — but that original person, they were creating content, put in actual money, dollars, and equipment.”
Mick Feldman, director of client services for The Now, an influencer marketing agency, said swapping one creator’s face for another without their consent was “completely unethical.”
He said the standard practice for agencies is to pay creators for the rights to use their content. It can cost tens of thousands to make a new video with a creator, but it typically costs a quarter of that to license an existing video from them, he added. Some creators have licensed their likenesses to companies that use them for AI ads.
It can be hard for creators to monitor if their videos are used for AI ads. They can add digital watermarks, but the practice is uncommon, Feldman said. Creators often don’t want to add a mark to their videos, and might not even consider themselves a likely target.
“Who would have thought that an apartment move-in video would be changed out like that?” he said.
