Attempts to avoid AI guardrails blamed for Swift deepfakes

Attempts to avoid AI guardrails blamed for Swift deepfakes

Time Stamp: February 5, 2024 5:11 PM
Attempts to avoid AI guardrails blamed for Swift deepfakes PlatoBlockchain Data Intelligence. Vertical Search. Ai.

The viral lewd deepfake images of popstar Taylor Swift reportedly stemmed from an online competition on 4chan, challenging contestants to break AI software content filters. 

Folks at Graphika, a New York social media analytics firm, have traced the pics directly to 4chan. People were asked to come up with tricks to bypass the safety guardrails on applications like Microsoft’s Designer or OpenAI’s DALL-E to create NSFW images of female celebrities. The pictures of Taylor Swift were created using Microsoft Designer, and were later shared in a group chat on Telegram, before being posted on X, according to Graphika.

“We are continuing to investigate these images and have strengthened our existing safety systems to further prevent our services from being misused to help generate images like them”, a Microsoft spokesperson told The Register in a statement. 

Deepfake images of Swift racked up millions of views on X before the social media platform removed them and temporarily blocked some search terms related to the songwriter. Graphika, however, said that the competition targeted other well-known female figures, and warned that the technology could be abused to target anyone. 

“While viral pornographic pictures of Taylor Swift have brought mainstream attention to the issue of AI-generated non-consensual intimate images, she is far from the only victim,” Cristina López G., a senior analyst at Graphika, told Bloomberg. “In the 4chan community where these images originated, she isn’t even the most frequently targeted public figure. This shows that anyone can be targeted in this way, from global celebrities to school children.”

The viral nature of the fake images portraying Swift in sexually suggestive poses, however, seems to have kicked lawmakers into gear. The White House’s press secretary Karine Jean-Pierre urged Congress to “take legislative action” in response to the “alarming” pictures. 

Last week, US senators introduced the Disrupt Explicit Forged Images and Non-Consensual Edits (DEFIANCE) Act of 2024, a bipartisan bill that would allow victims to sue people who create and share smutty AI-generated deepfakes. The legislation, spearheaded by Senators Dick Durbin (D-IL), Lindsey Graham (R-SC), Amy Klobuchar (D-MN), and Josh Hawley (R-MO), pointed to a study that claimed 96 percent of deepfake videos were pornographic and had been created with no permission or consent from the women it depicted. 

The DEFIANCE act was launched at the time top CEOs of social media platforms were questioned in a Congressional hearing over the sexual exploitation of children online, and lack of protection.

The Register has asked Graphika for further comment. ®

Time Stamp:

More from The Register