AI-generated images of Trump frequently appear! Taylor Swift trademarks her voice and appearance to prevent Deepfake.

To prevent AI deepfake images, Taylor Swift has applied for trademarks on her appearance and voice to combat infringement. At the same time, the United States is promoting the “Fake Disinformation Prevention Act,” which could fine illegal platforms up to $750k, hoping to improve intellectual property protection in the digital age through legislation.

Fighting AI Deepfake Fakes, Taylor Swift Applies for Voice and Appearance Trademarks

According to BBC reports, pop music superstar Taylor Swift has applied for trademarks on her voice and appearance to protect herself from AI-generated fakes. Her company submitted three trademark applications in the U.S. on April 24, including a stage photo from the Eras Tour and two promotional audio clips.

In recent years, AI-generated images of Taylor Swift have been widespread, ranging from explicit photos to fake ads calling for votes for Donald Trump. Some highly realistic images could cause public confusion and impact Taylor Swift’s personal image.

Image source: Gemini synthesis Fake AI-generated images on the internet show Taylor Swift supporting Trump, with Gemini adding the label “AI Fake Generated Image”

Trademark Law as a New Defense, Lawyers Say It Can Combat Confusingly Similar Marks

Trademark applications have become an emerging way to fight AI infringement. Before Taylor Swift’s actions, famous actor Matthew McConaughey became the first celebrity to use trademark regulations to protect his voice and image from AI misuse.

According to the trademark application documents, a photo representing Taylor Swift’s image shows her on stage holding a pink guitar, wearing colorful reflective tight clothing and silver boots. Additionally, she applied for trademarks on phrases like “Hey, I’m Taylor Swift.”

Image source: U.S. Patent and Trademark Office Trademark image of Taylor Swift’s appearance

Trademark attorney Josh Gerben stated that registering trademarks can effectively prevent AI misuse of images and voices.

He explained that registering specific phrases not only challenges reproductions of the same but also legally contests confusingly similar imitations. If someone creates an AI version of Taylor Swift wearing tight clothing and holding a guitar in the future, she would have the right to file a federal trademark claim.

U.S. Lawmakers Push Legislation to Address Deepfake and AI Infringement

While celebrities are applying for trademarks for self-protection, legislative bodies are also accelerating the construction of legal defenses.

U.S. Senator Marsha Blackburn announced on April 22 this year that she and another senator, Peter Welch, held a roundtable during the Recording Academy initiative period, with over 20 artists supporting bipartisan efforts for the “NO FAKES Act” and the “TRAIN Act.”

The purpose of these two bills is to protect creators from deepfake technology and ensure that copyrighted works are not illegally used to train AI models. Among them, the “NO FAKES Act” will establish the United States’ first federal public disclosure right, allowing individuals to control their digital avatars and pursue infringement liabilities.

The Fake Disinformation Prevention Act Regulates Digital Avatars, Fines Up to $750k for Violators

The “NO FAKES Act” grants individuals exclusive authorization rights, prohibiting unauthorized public display or distribution of computer-generated digital avatars, with a maximum protection period of up to 70 years after death. The bill includes reasonable use exemptions for news reporting and academic research, but explicitly excludes explicit sexual content.

In terms of penalties, the bill sanctions infringement through civil lawsuits. Violators face damages of $5,000 per work, and organizations must pay $25,000; online platforms that fail to cooperate with safe harbor mechanisms to remove infringing content could face fines of up to $750k.

AI Deepfake Fakes of Celebrities Could Also Pose Fraud Risks. Deloitte’s forecast indicates that by 2027, generative AI could cause fraud losses in the U.S. to reach $40 billion, a significant increase from $12.3 billion in 2023.

Although both bills are still in progress and have not yet been officially passed, the combined efforts of singers, actors, and other celebrities in trademark strategies signal that the industry and lawmakers are actively working to establish more comprehensive intellectual property protections to address emerging challenges in the digital era.

Further reading:
The songs I write are not mine! Copyright issues of Taylor Swift, Wu Qing-feng, G.E.M.: Can blockchain make royalty distribution fairer?
Countering AI Fake Photos! The iPhone camera app “ZCAM” launches, using encryption technology to reclaim authenticity