The internet was shocked this week by the appearance and rapid viral spread of explicit AI-generated images purporting to show global pop icon Taylor Swift nude. While Swift herself did not publicly acknowledge the existence of these nonconsensual deepfakes, her loyal fanbase took it upon themselves to report the content until it was removed by the platform it was being shared on. This incident has brought renewed scrutiny to the alarming pace at which new AI tools are being used to create and spread digitally altered content intended to humiliate and harm.
On Wednesday evening, an account on the popular video platform X posted a compilation of AI-generated nude images of Swift to their feed. The images appeared to show the singer fully nude in a football stadium, referencing the recent backlash Swift has received for attending NFL games to support her partner Travis Kelce. This content quickly went viral, amassing over 27 million views and 260,000 likes within 19 hours before the account was permanently suspended.
While the exact origins of the images are unclear, they seemed to have telltale signs of being AI-generated. A watermark on the images indicates they may have come from a known website that publishes fake nude celebrity imagery using AI deepfake technology. When scanned by Reality Defender, an AI detection software company, the images were assessed to have a high probability of being AI-created rather than real photos.
Regardless of their origins, the images represented a disturbing new evolution in the ongoing phenomenon of deepfakes – digitally altered video or image content created using AI. While many deepfakes are relatively harmless satire or entertainment, nonconsensual sexualized deepfakes have become an emerging issue as the technology behind them has advanced. However, most social platforms have lagged in developing policies or tools to effectively detect this kind of abusive deepfake content.
In Swift’s case, her supporters took matters into their own hands, launching a mass reporting campaign against accounts sharing the explicit images. Some who participated shared proof of notifications from X informing them that accounts had been suspended for violating policies against abusive behavior. Hashtags like “Protect Taylor Swift” trended as her fan base flooded related hashtags with positive messages about the singer.
While the swift action of her fanbase to report the deepfakes likely played a key role in taking them down, their ability to spread so rapidly in the first place has raised concerns. Taylor Swift has a particularly large and active online fanbase, but most victims of deepfakes do not have the same resources to counter them. The lag in developing tech-based solutions against deepfakes on major platforms has put the onus on victims to tirelessly report offending content.
Experts like Carrie Goldberg, an attorney specializing in sexual exploitation cases, assert that tech companies already have the means to get ahead of the issue. “AI on these platforms can identify these images and remove them,” Goldberg stated. “If there’s a single image that’s proliferating, that image can be watermarked and identified as well. So there’s no excuse.”
But in absence of more robust preventative measures by tech companies, the propagation of nonconsensual deepfakes continues to degrade and endanger women and girls especially. A 17-year-old actress recently found deepfakes of herself on X which the platform failed to remove. A 2023 NBC News investigation uncovered widespread nonconsensual deepfakes on X of underaged female TikTok stars.
For activists and legislators, Swift’s case underscores the need for definitive legal action. Deepfakes of this nature already violate policies on platforms like X, but policies clearly aren’t enough. A bill introduced in Congress last May by Rep. Joe Morelle would make nonconsensual sexual deepfakes a federal crime. But so far, it has seen no movement despite the urgency.
“Yet another example of the destruction deepfakes cause,” Morelle tweeted about the Swift incident.
This late-breaking case draws visceral attention to the possible damage enabled by ever-more-accessible AI generative models. While lawsuits and legislation play catch up, the onus remains on platforms to refine their approach beyond after-the-fact moderation. For victims like Taylor Swift, their power still lies mainly in the ability to mobilize a community for reporting and takedowns. For now, all anyone can do is guard their own reputation and digital presence as new threats continue emerging.