Why FKA twigs created a Deepfake AI version of herself

BBritish singer-songwriter FKA twigs, born Tahliah Debrett Barnett, testified before the U.S. Senate Judiciary Subcommittee on Intellectual Property on Tuesday about the dangers of artificial intelligence. She reported that she was especially concerned as she is an artist whose music and performances are used by third parties to train artificial intelligence models. She said the power of this technology became especially apparent to her as she tried to build a fake version of herself.

“Last year, I developed my own deepfake version of myself, which is not only trained on my personality, but can also use my exact tone of voice to speak multiple languages,” the singer said in her statement. “I will be engaging my ‘AI branches’ later this year to broaden my reach and handle my online interactions on social media, while continuing to focus on my art from the comfort of my studio.”

FKA twigs emphasized that in certain cases, deepfake technology can be useful and practical, as long as it is used with the consent of the person it is imitating. “These and similar emerging technologies are highly valuable tools, both artistically and commercially, when they are under the control of the artist. What is not acceptable is when my art and identity can simply be taken by third parties and falsely exploited for their own gain without my consent, due to the absence of appropriate legislative control,” she argued.

The musician especially encouraged those in power to enact legislation to control deepfake technology and warned them not to repeat the mistakes made when the Internet was invented. “Three decades ago, we did not realize that the Internet would become so deeply integrated into the core of our daily lives. Policies and controls to keep up with the emergence of technology have not been implemented to protect artists, young people, those who are vulnerable,” she said.

FKA twigs is not the only person to express concerns about the use of personal intellectual property by generative AI companies. On Tuesday, a group of eight American newspapers announced that they would be suing Microsoft and Open AIcreator of ChatGPT, for using his work as training data without consent.

The US does not yet have any federal law that criminalizes the use of deepfake technology without the permission of the person being impersonated. This has been particularly harmful to people in abusive relationships, who increasingly become victims of revenge porn using deepfake technology.

At the beginning of this year, Fake pornographic images of Taylor Swift have circulated widely on social media platform X (formerly Twitter), highlighting the need for new legal protections. Content moderators removed them, bringing the matter to public attention.