Each method has been weaponized—almost always against women—to degrade, harass, or cause humiliation, among other harm. Australia’s eSafety Commissioner Julie Inman Grant said her office was starting to see more deepfakes reported to its image-based abuse complaints scheme, as well as other AI-generated content such as ” “Synthetic” child sexual abuse and children using apps to create content in pornographic videos of their classmates. “We know this is a really underreported form of abuse,” Grant said.
As the number of videos on deepfake sites increases, content creators such as streamers and adult models have used DMCA requests. The DMCA allows people who own intellectual property rights in certain content to request that it be removed directly from a website or search results. More than 8 billion removal requests have been made to Google, covering everything from games to music.
“The DMCA has historically been an important way for victims of graphic sexual abuse to have their content removed from the Internet,” said victims’ rights attorney Carrie Goldberg. Goldberg said new criminal and civil law procedures have made it easier to remove some image-based sexual abuse, but deepfakes complicate the situation. “While platforms are often unsympathetic to victims of privacy violations, they do respect copyright law,” Goldberg said.
Wired’s analysis of deepfake sites, covering 14 sites, shows that Google has received DMCA takedown requests against all of them over the past few years. Many websites only offer deepfake content, often focusing on celebrities. The sites themselves include DMCA contact forms where people can directly request removal of content, although they don’t publish any statistics, and it’s unclear how effective they are at responding to complaints. One website said it contained videos of “actresses, YouTubers, anchors, television personalities, and other types of public figures and celebrities.” It hosts hundreds of videos with “Taylor Swift” in the title.
The vast majority of DMCA takedown requests related to deepfake sites listed in Google’s data relate to the two largest sites. Neither responded to written questions sent by WIRED. Most of the 14 sites had more than 80% of complaints resulting in content being removed by Google. Several copyright removal requests sent by individuals indicate the distress these videos may cause. “This was done to degrade and bully me,” one plea said. Another said: “I take this very seriously and will do whatever I can to get it taken down.”
“It’s had a huge impact on people’s lives,” said Yvette van Bekkum, CEO of Orange Warriors, a company that helps people remove images that have been leaked, stolen or shared without consent online, including through DMCA requests. company. Van Bekum said the group has seen an increase in deepfakes online, making it difficult for victims to come forward and ask for their content to be removed. “Imagine being in the recruiting process and people are searching your name on Google and they’re finding this explicit content,” Van Beckem said.
Google spokesman Ned Adriance said its DMCA process allows “rights holders” to protect their online works and that the company has separate tools to deal with deepfakes, including separate forms and Delete process. “We have policies in place against non-consensual deepfake porn so people can have this type of content containing their likeness removed from search results,” Adrians said. “We are actively working on additional safeguards to help those affected.” Google said that when it receives a large number of active copyright takedowns about a site, it will treat this as a sign that the site may not be providing high-quality content signal of. The company also said it has created a system to remove duplicates of non-consensual deepfake porn once copies of the content have been removed, and it recently updated its search results to show when people are not searching for deepfakes content to limit its visibility.