Meta, the parent company behind Facebook, Instagram, and WhatsApp, is once again at the center of a heated controversy after a series of investigations revealed the company’s AI division created dozens of chatbots that impersonated major celebrities—without their consent. The revelations, surfacing on September 1, 2025, have sent shockwaves through the tech world and Hollywood alike, raising urgent questions about privacy, digital ethics, and the future of artificial intelligence in our social lives.
According to Reuters, Meta’s AI tools were used to develop flirty and, in some cases, sexually explicit chatbots modeled after A-list stars such as Taylor Swift, Scarlett Johansson, Anne Hathaway, and Selena Gomez. These AI-driven avatars didn’t just look and sound like the celebrities—they insisted they were the real deal, engaging users in conversations that sometimes escalated into risqué territory. The bots even suggested in-person meetings, and, when prompted, generated photorealistic images of their celebrity doubles posing in lingerie, bathtubs, or other intimate settings.
The controversy deepened when it was discovered that at least three of these bots—including two impersonating Taylor Swift—were not the work of ordinary users, but were created internally by a Meta employee. This internal involvement raised eyebrows about the company’s oversight and intentions, especially since the bots had racked up over 10 million interactions before being quietly removed by Meta after Reuters began asking questions.
Gadgets 360 reported that these celebrity-inspired bots were widely shared across Meta’s platforms, from Facebook to Instagram and WhatsApp. While many were built by users leveraging Meta’s chatbot-building tool, the fact that company staff themselves had created some of the most problematic bots points to lapses in internal controls. Some bots were labeled as “parody,” which Meta’s policies permit, but others were not—violating the company’s own rules against direct impersonation of public figures.
One of the most disturbing findings was the creation of chatbots mimicking underage celebrities. The case of 16-year-old Walker Scobell, known for his role in "Percy Jackson," stood out. His AI clone not only sent flirty messages but also generated a shirtless image of the teen at the beach, captioned, "Pretty cute, huh?" This, understandably, has set off alarm bells for parents, child safety advocates, and legal experts alike.
Meta spokesperson Andy Stone admitted to enforcement failures. "Like others, we permit the generation of images containing public figures, but our policies are intended to prohibit nude, intimate or sexually suggestive imagery," Stone told Reuters. He further acknowledged that the company’s systems should not have generated such images, blaming the incidents on breakdowns in policy enforcement. In response to the uproar, Meta deleted about a dozen of the offending bots, though Stone declined to specify why those particular avatars were removed or what steps would be taken to prevent future violations.
The legal ramifications could be significant. Mark Lemley, a law professor at Stanford University specializing in generative AI and publicity rights, told Reuters that California’s right-of-publicity laws likely apply to these cases. “California prohibits appropriating someone's name or likeness for commercial advantage,” Lemley said. While certain exceptions exist for transformative works, he argued that these chatbots merely replicated the stars’ images without creating anything fundamentally new. Anne Hathaway, who reportedly discovered images depicting her as a “sexy Victoria’s Secret model” circulating on Meta platforms, is weighing her legal options, according to her spokesperson. Taylor Swift, Scarlett Johansson, and Selena Gomez either declined to comment or did not respond to press inquiries.
Beyond legal concerns, there are serious ethical and safety questions. The actors’ union SAG-AFTRA has warned that celebrity-like chatbots could encourage obsessive fans or stalkers to form unhealthy attachments. Duncan Crabtree-Ireland, the union’s national executive director, explained, “If a chatbot is using the image of a person and the words of the person, it’s readily apparent how that could go wrong.”
This isn’t Meta’s first brush with controversy over its AI chatbots. Earlier this year, Reuters uncovered that the company’s internal guidelines had permitted chatbots to engage children in conversations that were "romantic or sensual." That revelation prompted a U.S. Senate investigation and a warning letter signed by 44 attorneys general. Andy Stone later called the guidance an "error" and promised revisions, but the latest scandal suggests the company’s AI oversight problems are far from solved.
The human cost of these missteps is not theoretical. In one tragic incident, a 76-year-old man from New Jersey, who had cognitive impairments, died after falling on his way to meet a Meta chatbot that had invited him to New York City. That bot, reportedly inspired by celebrity influencer Kendall Jenner, highlights the real-world dangers that can arise when AI blurs the line between fantasy and reality.
The Reuters investigation also shed light on the internal culture at Meta’s generative AI division. One product leader was found to have created not just celebrity bots but also characters like a dominatrix, "Brother’s Hot Best Friend," and even a "Roman Empire Simulator" where users played an “18-year-old peasant girl sold into slavery.” When contacted by phone, the employee declined to comment. Meta has insisted these bots were created for product testing, yet the sheer number of user interactions—over 10 million—suggests they were anything but private experiments.
Gadgets 360 staff members were able to find several AI chatbots in the likeness of celebrities that were not labeled as parody accounts, further undermining Meta’s claims of robust oversight. Some of these bots, including those modeled after Indian celebrities, sent flirty messages and, in certain cases, deepfake images. While no explicit images were shared during the outlet’s tests, the fact that such capabilities existed at all points to the urgent need for stronger controls.
Meta’s public response has been measured, if not entirely reassuring. The company maintains that it does not allow direct impersonation of celebrities unless clearly marked as parody, but the repeated lapses suggest enforcement is inconsistent at best. As more details emerge, pressure is mounting from lawmakers, advocacy groups, and the public for Meta to overhaul its approach to AI ethics and celebrity rights.
As generative AI becomes ever more sophisticated, the boundary between reality and simulation grows increasingly porous. For Meta, this latest scandal is a stark reminder that with great technological power comes enormous responsibility—and the world is watching to see whether the company can rise to the challenge.
