Can you trust a celebrity’s endorsement? The question is no longer just about whether you trust their judgment; it’s about not knowing if they ever really made the endorsement in the first place.
As AI models generate increasingly hyper-realistic images and videos, it’s causing an explosion of deepfakes and impersonation scams involving public figures. Global deepfake scammers made off with $96 million in just the first three months of 2026, with the majority of successful scams hinging on imitating government officials or celebrities to endorse fraudulent investment opportunities, according to SurfShark.
Now, a new startup, PersonaShield, is launching to not only help public figures stamp out deepfake scams, but take greater control of their likeness overall in this new AI world. The company is building infrastructure around “AI identity rights,” helping creators, influencers, celebrities, executives and public figures manage how their image and likeness are used across AI systems and social platforms.
“Every creator we've spoken to has said, ‘well, I want people to be able to generate images of me. I just have certain requirements,’” said founder and CEO Phillip Shoemaker.
PersonaShield’s service allows public figures to define what uses of their likeness are permitted, automate enforcement around misuse, and even create pathways to monetize their likeness through authorized uses like fan art and online content. For example, an executive could ban all uses of their likeness, while an influencer could set terms to allow their likeness to be used in certain ways for a licensing fee.
“We give guidance, but everything's up to them to choose how they protect it, if they want to protect it, and if they want to monetize it,” said Seamus Hennessy, product manager at PersonaShield.
There are already various startups focused on detecting deepfakes, a crowded field of take-down services, and some creative solutions for designating authorized content. With PersonaShield, Shoemaker is looking to combine both sides of public figures’ modern identity concerns, bringing together protection and marketplace.
“We're about owning your likeness, and protection is one aspect of it,” he said. “But image generation is another aspect — image generation with monetization and paying the people whose images you're generating from.”
Choose your own (likeness) adventure
As Apple’s former head of app review who helped create app store guidelines, and later an executive and advisor in the decentralized identity space, Shoemaker knows a thing or two about managing scams and fraud online. He also knows about how quickly enthusiasm to monetize on a new technology opportunity can rise.
In addition to servicing public figures, PersonaShield also serves as a marketplace. People can go to the PersonaShield website, browse and search for celebrities or influencers, and purchase permission to use a public figure’s likeness however they want — so long as the use falls within the person’s set guidelines. This isn’t for using their likeness in, say, a feature film, but rather for fan engagement, online content, memes, and the like.
Image generation sample for skateboarder Bob Burnquist. Credits: PersonaShield
PersonaShield’s first customer was professional skateboarder Bob Burnquist. Shoemaker described Burnquist as making it very clear that if he’s in a skate competition, he needs to have Banco do Brasil on his shirt. Because it affects his sponsorship, it’s his absolute requirement, and he can't have people generating images of him without it.
Combating AI with AI
For PersonaShield, AI is the problem and also a large part of the solution. Like most startups, it’s using AI widely throughout the company for everything from engineering to marketing. What’s more, AI is also a core part of the company’s service and technology infrastructure underneath its image analysis capabilities.
“We use AI to tell us, is this an image of Snoop Dogg? And if so, what's he smoking? Oh, it's not cannabis; it's crack. Okay, that falls outside of his safeguards,” said Shoemaker as an example (Snoop Dogg is not currently a customer).
Hennessy compared the current methods for deepfake detection used throughout the industry to a game of whack-a-mole, largely dependent on analyzing pixels to classify images as deepfakes in a more strict, deterministic way. With PersonaShield, he said they’re trying to bring more context and understanding about the specific person into the detection pipeline. This includes analyzing what they’re known for, what they’re often associated with, and what’s counter to that, and overall, building a profile of the person and using that to flag content that falls outside their set safeguards.
“We're trying to think about, how can we make this more adaptable? So no matter how accurate these images get, it can almost think for itself in terms of like, I know who this creator is, what content they promote, what they tend to stand for, what they say, who their sponsors are, where they’re known to be, and what they tend to promote or not,” Hennessy said.
PersonaShield’s ambition for the next few months is to use AI to automate takedown requests for unauthorized content. This is currently an incredibly tedious task, which involves manually reaching out to each platform and requesting content to be removed (and following up and escalating if it’s not). The idea is for agents to handle it on the backend without any manual interaction.
PersonaShield founder and CEO Phillip Shoemaker. Credits: PersonaShield
Speaking about the rising deepfake issue, Variety reports that Hollywood agents (the human kind, who represent celebrities) say the previous manual methods for pursuing takedowns are insufficient for the sheer number of violating posts they’re seeing because of AI. Agency sources describe meeting with “dozens” of startup solutions and told Variety they expect — and are already seeing — more talent lean into these kinds of automated, all-in-one solutions. Other choices include Loti and Vermillio, which also help facilitate licensing opportunities and scans the internet for unauthorized content to automatically issue takedown requests.
Overall, PersonaShield wants to put people in the driver’s seat in this new AI era. To the team, this means empowering people who feel like the problems associated with AI and deepfakes were thrust upon them without any say. As Shoemaker sees it, the frustration and anxiety come from not having a choice in any of this.
“So many people are angry with [AI] because it's all trained on our data without even asking us,” he said. “We want to move forward and start asking the questions, like, do you want to allow images to be generated of you? What do you want generated, and what don't you want generated?”
Top photo: Takedown request sample for skateboarder Bob Burnquist. Credits: PersonaShield
