(NewsNation) — Google is making changes to its search engine in an effort to crack down on nonconsensual deepfake pornography that has exploded with the rise of generative artificial intelligence.
The company said Wednesday it has developed a new system that will make it easier for victims of deepfake porn to take down photos and videos of themselves.
For years, people have been able to request that nonconsensual content be removed, but now, Google will filter all explicit results on similar searches and take down any duplicates.
“These efforts are designed to give people added peace of mind, especially if they’re concerned about similar content about them popping up in the future,” Google said in a blog post.
Google is also tweaking its ranking system to make it harder for people to find explicit fake content and demoting sites that have received a “high volume of removals” for deepfake pornography.
For example, someone searching for a deepfake image of Taylor Swift will instead be served with “high-quality, nonexplicit content” like “relevant news articles.”
Google said previous updates have already cut explicit image results on those types of queries by over 70%.
The company said it’s also working on improvements so its search engine can better differentiate between real explicit content, like an actor’s nude scene, and fake explicit content.
The rise of easy-to-use generative AI tools has led to a surge of deepfakes in recent years, a type of synthetic media that makes it possible to overlay someone’s face onto a naked body. Around 96% of deepfakes are nonconsensual pornography, according to one study.
It’s a problem major tech companies are trying to solve.
In May, Google banned advertisers from promoting services that can be used to generate deepfake porn.
Another Silicon Valley giant, Meta, has also acknowledged it needs to do more to combat deepfake pornography. Last week, the company’s oversight board determined that Meta’s rules prohibiting “derogatory sexualized photoshop” are not “sufficiently clear” to its users.
The board made several suggestions for tweaking the language in its rules. Meta said it would review the board’s recommendations.