GitHub’s Deepfake Porn App Still Not Working

“When we look at intimate image abuse, the vast majority of tools and weaponized use come from the open source space,” says Ajder. But they often start with well-intentioned developers, he says. “Someone creates something they think is interesting or cool, and someone with malicious intent recognizes its harmful potential and weaponizes it.”

Some, like the disabled depot in August, have purpose-built communities around them for open use. The model has positioned itself as a vehicle for deeply fake porn, which Ajder claims is becoming a “funnel” for abuse, mainly targeting women.

Other videos have been uploaded to a porn streaming site The AI ​​models, downloaded from GitHub, rendered the faces of famous deepfake targets Emma Watson, Taylor Swift and Anya Taylor-Joy, as well as other lesser-known but very real women in sexual situations, according to the account.

The developers have freely described the tools they use, including two that have been cleaned up by GitHub but whose code is kept in other existing repositories.

Deepfake-seeking criminals congregate openly in many places online, including secret community forums on Discord and Reddit, boosting efforts to prevent deepfakes. On September 29th, a Redditor offered his services using the software from the archived repository. “Could anyone do my cousin,” he asked.

Torrents of the main repository, banned by GitHub in August, are also available in other corners of the internet, showing just how tough the police are. open source deepfake software across the board. There have been other deep fake porn tools like the DeepNude app lowered in the same way before new versions appear.

“There are so many models, so many different forks in the models, so many different versions, that it can be hard to keep track of them all,” says Elizabeth Seger, director of digital policy at the UK cross-party think tank Demos. “Once a model is open for download with open source, there is no way to bring it back to the public,” he adds.

One deeply fake porn creator, with 13 manipulated nude videos of female celebrities, considers a prominent GitHub repository marketed as a “NSFW” version of another project that promotes responsible use and explicitly asks users not to use it for nudity. “Learn all available Face Swap AI from GitHUB, without using online services,” their profile on the tube site says cheekily.

By the time WIRED identified the deeply fake videos, GitHub had already disabled this NSFW version. But other depots, known as “unlocked” versions of the model, were available on the platform on January 10, including one with 2,500 “stars”.

“Technically it is true that once (a model is in place) it cannot be undone. But we can still make it harder for people to access,” Seger said.

Left unchecked, he adds, the potential for deeply fake “porn” harm isn’t just psychological. Its powerful effects include the intimidation and manipulation of women, minorities, and politicians. deep political frauds influencing women politicians globally.

But it’s not too late to get the problem under control, and platforms like GitHub have options, Seger says, including intervention at the upload point. “If you put a model on GitHub and GitHub says no, and all the hosting platforms say no, it becomes difficult for a normal person to get that model.”

Politicians, tech companies, developers, and of course creators of offensive content also rely on controlling deeply fake porn made with open source models.

At least 30 U.S. states have some sort of deep-fake pornography laws, including bans, according to the nonprofit Public Citizen’s. legislative followeralthough definitions and policies vary, and some laws only cover minors. Deepfake creators in the UK will also feel the force of the law following the government’s announcement criminalizes the creation of deep forgeries with sexual contentas well as their sharing, on January 7.