Nobody is safe from celebrity AI deepfakes

An AI-generated image of actress Mia Goth wearing a shirt reading "MY EX IS MY BIGGEST FAN" spread widely on Instagram after being posted by a fan account with over a million followers and a blue verified checkmark. Goth's representatives said they had previously complained to Instagram about counterfeit AI images of her. Meta permits users to purchase verified checkmarks even for non-official accounts, and still images are exempt from its AI labeling requirement, meaning the fake Goth image needed no disclosure. The post spread to celebrity gossip outlets and Google search results, where an AI-generated overview repeated the false claim that Goth was mocking her ex-husband Shia LaBeouf. Instagram eventually removed the account for impersonating a public figure.

The article documents a broader pattern of AI-generated celebrity deepfakes proliferating across social platforms. These range from overtly sexualized content featuring female celebrities like Scarlett Johansson, Millie Bobby Brown, Zendaya, and Ariana Grande, to viral videos of AI Taylor Swift rapping and AI-rendered Formula 1 drivers lipsyncing. Some accounts monetize this content through subscriptions, advertising, and online courses. After The Verge inquired about eight such accounts, Meta removed some, stating that using someone's likeness to pose as that individual is not permitted regardless of whether content is authentic or AI-generated.

In the absence of a US national right of publicity law, enforcement varies by platform. The Take It Down Act of 2025 criminalizes nonconsensual intimate imagery including AI-generated content. YouTube operates a likeness detection program that scans for visual and audio matches and allows takedown requests, removing over 95 percent of reported videos in early 2026. However, the overall volume of AI celebrity content keeps growing, and the article concludes that it is unclear whether anyone is trying or able to stop it.

(Source：The Verge)