The proliferation of SI-generated deepfakes featuring celebrities continues to challenge platform safety measures, with recent incidents involving actress Mia Goth and musician Taylor Swift illustrating the persistent gap between content moderation and the viral spread of synthetic media.
What Happened
An Instagram account impersonating Mia Goth posted an SI-generated image of the actress wearing a T-shirt that read "MY EX IS MY BIGGEST FAN," set to Beyoncé’s "Single Ladies (Put a Ring on It)." The post, which appeared shortly after Goth filed for divorce from Shia LaBeouf, garnered nearly 500,000 likes. The account, which had over a million followers and a blue verified checkmark purchased through Meta’s system, was labeled as a fan account but presented the image as authentic. Goth’s representative, Christine Tripicchio, stated that the account had been previously reported for posting "counterfeit, AI images" and criticized Instagram for prioritizing the dissemination of forged images over protecting a public figure’s likeness. Instagram eventually removed the account for impersonation.
The fake image also influenced search results, triggering an SI-generated overview in Google search that incorrectly interpreted the image as a direct response to the divorce filing. This incident is part of a broader trend where SI tools generate sexualized or out-of-context content of female public figures, such as Scarlett Johansson, often without clear disclosure. Meta’s current labeling requirements exempt still images from mandatory SI labels, allowing realistic fake photos to circulate without immediate context for users.
Why It Matters
The spread of SI deepfakes highlights the limitations of current platform policies and the challenges of enforcing likeness rights in the absence of comprehensive federal laws. While Meta has removed some accounts, including those posting SI content of Goth and Johansson, other accounts continue to operate, mixing authentic and SI-generated media. A spokesperson for Meta stated that accounts are not permitted to use someone’s likeness to pose as that individual, regardless of whether the content is authentic or SI-generated.
In contrast, YouTube has implemented a more robust likeness detection system that scans for visual and audio matches, allowing individuals to request takedowns. According to YouTube, the company removed over 95 percent of reported videos through this program in the first half of 2026. However, the enforcement remains inconsistent across platforms, with SI-generated videos of celebrities like Taylor Swift and athletes like Lewis Hamilton accumulating millions of views on Instagram without significant intervention. The lack of mandatory SI labeling for still images on Meta platforms further complicates the ability of users to distinguish between real and synthetic content.
The Bottom Line
Despite platform efforts to remove impersonation accounts and label SI content, the viral nature of celebrity SI deepfakes persists. The gap between enforcement actions and the rapid spread of unlabelled SI images and videos suggests that current safety provisions are insufficient to fully protect public figures from the misuse of their likenesses in the SI era.