Bryce Gruber was scrolling through TikTok when a self-tanner promotion appeared on her For You page. The woman rubbing the product onto her body looked very familiar.
It was her. Sort of.
“This is super scary and weird,” Gruber, an editor and writer behind the Substack, Extra Good, said in a TikTok posted Aug. 5 “By the way, this is not me. I mean, it’s me, but it’s robot me,” she told her 193,000 TikTok followers.
The video Gruber encountered featured an AI-generated version of her enthusiastically reviewing a self-tanner: “This tanner changed my entire routine,” her digital doppelgänger says in the clip.
As the AI version of Gruber rubs the product across her body, the real Gruber says that she had never seen or used the product before and had no relationship with the company selling it (though she didn’t identify the brand).
“Obviously, I freaked out over this,” she said. “I never saw that product in my life before, and I probably never want to.”
The strange encounter led Gruber back into her TikTok settings, where she discovered just how much access other users appeared to have to her digital double.
How does TikTok AI Cast work?
The feature behind the video is called AI Cast, which lets users create an “AI Self” that can appear in AI-generated videos. TikTok’s Help Center describes AI Self as one of the “digital avatars or voices” users can create from their face and voice information.
This Tweet is currently unavailable. It might be loading or has been removed.
Unlike a traditional deepfake made by scraping someone’s existing posts, creating an AI Self requires a user to deliberately set one up.
TikTok requires users to verify their identity before creating an “AI Self,” a source at the company told Mashable. The process involves recording themselves from multiple angles and reading a series of numbers after selecting “Add AI Self” from their profile.
When asked about Gruber’s case, the source said, “This creator would have had a notification that [someone] was using AI Cast.” Users can also control who has access to their “AI Self,” keeping it private or sharing it with friends or the public. TikTok says anyone whose AI Self appears in another person’s video can delete the video or draft directly, regardless of who created it. Once they do, the original creator can no longer access the content.
However, once that digital version exists, AI Cast can put it into scenarios that the real person never filmed. A user can generate a video starring their own AI Self, or bring another person into the video by @-mentioning a friend or user whose AI Self is available to them. In other words, two people don’t have to get together to film something for their digital versions to appear together in a video.
According to Gruber, though, when she opened up her own AI Self settings, she said her permissions were set to “Everyone,” which she described as the default. She urged users to switch the setting to “Only You.”
“If it’s on ‘Everyone,’ you become a self-tanner ad,” she joked.
Notably, Gruber told Mashable she reported the self-tanner video to TikTok, but was told that because the AI-generated voice differed from her own, the platform did not consider the person depicted in the video to be her. She isn’t sure whether the brand faced any consequences.
Mashable Trend Report
And there were differences. Her voice and natural cadence weren’t quite right, she said, while other details had seemingly been invented altogether.
“My voice and speech cadence are definitely different in the AI video, but also little details like the freckles on my arms and legs are very different — and the manicure?” she said. “I haven’t had a French manicure since 1999, so that was a funny touch to see.”
AI doubles are becoming big business
Platforms and advertisers are increasingly experimenting with synthetic influencers.
TikTok itself has been building AI advertising tools through its Symphony platform, which lets advertisers generate videos using AI avatars. In June, the company expanded those tools through Symphony Agent, which can produce ads using generative video, scripts, and avatars.
The technology has also advanced to the point where creators don’t necessarily need to build an official digital twin for someone to imitate them. In Mashable’s 2025 investigation into AI-generated YouTube ads, UC Berkeley professor Hany Farid explained just how little source material is required.
“With just 20 seconds of a person’s voice and a single photograph of them, it is now possible to create a video of them saying or doing anything,” Farid told Mashable’s Technology editor, Timothy Back Werth.
Gruber says she has already experienced that version of the problem, too.
After finding the self-tanner video, Gruber said she discovered that a Chinese denim company had previously used her face and voice, and even recreated what looked like her backyard, to market jeans. According to Gruber, the company sold more than 1,000 pairs.
She is far from the only creator confronting that problem. In June, guitarist and creator Sophie Burrell told Guitar World that she stumbled across a TikTok that reproduced one of her videos almost exactly — down to the camera angle, room, guitar, and movements. YouTube musician Rhett Shull also discovered an AI-generated guitar channel that appeared to recreate pieces of his older videos, including his former filming set. The channel, named “Guitar Gems With Chase,” had accumulated more than 5,000 subscribers and 600,000 views when Shull found it.
Plus, as Mashable reported last year, the problem was already showing up in advertising: AI-generated videos impersonated Hoda Kotb to promote weight-loss products and Elon Musk in crypto scams, while synthetic financial advisers and even fabricated small-business owners appeared in deceptive YouTube ads.
When asked whether brands can use someone’s AI Self to create promotional content, the TikTok representative said the company would “look into it.”
“It should just be more of a peer-to-peer kind of creator-level product,” the source said.
There are rules against at least some of this behavior. The Federal Trade Commission’s rule on fake reviews and testimonials prohibits testimonials that falsely represent someone as having experience with a product they never used.
The FTC has even addressed AI advertising specifically. In guidance that mentions TikTok’s Symphony Avatars by name, the agency says AI avatars aren’t inherently prohibited in advertising, but their use could violate federal rules when they create a fake or misleading testimonial.
“Just last week, we announced that we’re testing a new feature that helps folks find accounts on TikTok who are maybe using their likeness,” a source at TikTok told Mashable. “You could easily upload your photo and verify your ID through the third-party platform, and it will then confirm you’re the owner of your identity.”
“We are testing strategies broadly to help protect AI likeness on the platform,” he added.
Instagram already learned how sensitive this is the hard way
TikTok isn’t the only platform discovering how quickly users react when control over their likeness becomes an opt-out setting.
Last month, Meta launched a feature for its new Muse Image generator that allowed people to @-mention public Instagram accounts and use their photos as references for new AI-generated images. Users could disable this feature in their settings, but it was enabled automatically for public accounts.
This Tweet is currently unavailable. It might be loading or has been removed.
The reaction was swift. SAG-AFTRA criticized the opt-out approach, while Emmy-winning actor Hannah Einbinder urged her followers to disable the setting. Three days after introducing it, Meta pulled the feature entirely.
“Our intent was to provide a useful creative tool and to give people control over whether their public content could be referenced in this way,” Meta said in an update. “We’ve heard the feedback that this feature missed the mark, so it’s no longer available.”
The episode gets at the same question now raised by Gruber’s TikTok experience: If your AI self can say and do things you never did, where does your control over your own likeness begin…and end?
For now, Gruber has considerably more practical advice: “Careful out there with this stuff,” she said. “It can happen to anyone.”
Topics
Artificial Intelligence
Creators
