A fake Instagram photo of actress Mia Goth wearing a shirt that read “MY EX IS MY BIGGEST FAN” drew nearly 500,000 likes within days. It was entirely AI-generated, and it was posted by an account with more than a million followers and a blue “verified” checkmark. The image appeared just after Goth filed for divorce from actor and filmmaker Shia LaBeouf, and it was designed to look like a jab at him.
The account, @miagoth_, was labeled as a fan page. That didn't stop peers in the entertainment industry from engaging with the post. Goth's representative, Christine Tripicchio, told The Verge that the account had qualified for a blue check, which she called “disconcerting on a fundamental level.” She said the account's operators had been reported to Instagram before for posting “counterfeit, AI images” of Goth.
Meta allows users to buy blue checkmarks, even when an account isn't run by a prominent figure or isn't an official page. In this case, the verification gave the fake image a credibility it did not deserve.
The post spread well beyond Instagram. Because of the divorce news, celebrity gossip sites picked up the image and wrote about it, and that coverage fed into Google search results. A search for Goth or the shirt returned an AI-generated overview saying the actress was taking “a clear swipe at her ex, Shia LaBeouf, following their recent divorce filing.” The overview cited the blogs that had covered the post. A search result headlined with the fabricated quote also appeared prominently, with links to social media posts about the photo. Instagram eventually removed the account for impersonating a public figure.
The Goth case is one example of a much larger problem. Famous people have been a frequent target of consumer AI tools since they became widely available several years ago. An AI-generated Drake sound-alike song was pulled from YouTube after copyright strikes. Scarlett Johansson hired a legal team after OpenAI released a ChatGPT voice that sounded like her, and she did not consent to it. Ariana Grande reportedly blocked an account that posted AI-generated content of her last year. Taylor Swift publicly endorsed Kamala Harris for president in 2024, saying she felt compelled to clarify her position after Donald Trump posted fake images suggesting she had endorsed him.
Content involving female public figures is often sexual. A profile that posts only AI images of Johansson, frequently showing her bare feet, listed contact information for advertisers and linked to Patreon and Telegram channels that sell subscriptions. Another account posts realistic AI images of actors and musicians, including Millie Bobby Brown, Zendaya and Grande, nearly always partially undressed or in suggestive poses.
After The Verge asked Meta about eight accounts posting celebrity deepfakes, spokesperson Alex Dziedzan said the company had removed some of them. The Goth account and the Johansson account appear to have been taken down. “We do not permit accounts to use someone's likeness to pose as that individual, regardless of whether the content is authentic or AI-generated,” Dziedzan said in an email.
Not all of the content is malicious. Instagram is full of clips of AI-generated celebrities lip-syncing to popular songs. One shows Taylor Swift rapping along to a Drake track, and it has 1.8 million views. Others feature AI versions of Shrek, Steve Jobs, Lionel Messi and Formula 1 drivers Lewis Hamilton and Max Verstappen, some with millions of views. Many are obviously fake, but viewers often don't seem to mind.
Uneven rules across platforms
There is no national right of publicity law in the United States that governs AI likenesses, so disputes over AI-generated content of real people are mostly settled by the platforms themselves. The federal Take It Down Act, signed into law in 2025, makes it a crime to distribute nonconsensual intimate imagery, including AI-generated images, and requires platforms to remove it within 48 hours.
YouTube runs a likeness detection program that scans for visual matches of a person and, more recently, sound-alikes. People enrolled in the program can request takedowns. YouTube says it weighs public interest, newsworthiness and consent before acting on a request. The program started with public figures such as celebrities, creators, politicians and journalists, and it now covers every adult with a YouTube account. YouTube spokesperson Jack Malon said the system was developed with the entertainment industry and that the company removed more than 95 percent of the videos reported through it in the first half of 2026.
Meta's rules are looser. The company requires an AI label on realistic videos of people and on AI-generated voice audio. Still images are exempt, which means the fake Goth photo would not have needed a disclosure. Meta has also added “AI-generated profile” labels for accounts whose avatars depict people who do not exist.
Even after one account is taken down, others keep posting. A Johansson account with more than 300,000 followers remained live at the time of reporting. It posts a mix of real and AI-generated images, and its most popular AI image, which shows Johansson lying on a couch with bare feet in the foreground, has nearly half a million likes. The post carries no AI disclosure, which Meta's rules do not require for still images. Commenters seem to know the picture is not really her, but many don't seem to care.
Questo articolo è stato scritto con l'assistenza dell'IA.
News Factory APP - notizie agentiche per potenziare il tuo SEO e AEO.