AI can copy your favorite artist in seconds: Stopping it could take years in court

From fake songs to cloned celebrity vocals, AI is moving fast and the law is still trying to catch up.

AI voice cloning can be challenged in court, but celebrities do not own a simple nationwide copyright in the natural sound of their voices. According to the U.S. Copyright Office, a person’s voice falls outside the subject matter of copyright, even though a specific recording of that voice may receive copyright protection. The agency also concluded that existing laws leave significant gaps and recommended a nationwide law covering unauthorized digital replicas. 

That does not leave artists powerless. Decades before generative AI became available to the public, courts recognized that a famous voice could be part of a performer’s legally protected identity. In Midler v. Ford Motor Co., the Ninth Circuit ruled that deliberately imitating a widely recognized singer’s distinctive voice to sell a product could violate California law. The court did not declare every vocal imitation illegal. It focused on recognizable identity, deliberate copying and commercial exploitation. 

AI voice cloning may also create a federal false endorsement claim. In Waits v. Frito Lay, the Ninth Circuit recognized that an unauthorized imitation of a distinctive voice could violate the Lanham Act when consumers might believe the entertainer approved or participated in an advertisement. That theory becomes especially important when a cloned celebrity voice promotes a product, appears in sponsored content or is presented as an official song. 

State law can offer even stronger protection. Per the Tennessee General Assembly, the ELVIS Act added a person’s voice to the identity rights protected under Tennessee law and allows courts to restrain certain unauthorized uses. Per the California Governor’s Office, California’s AB 2602 requires performer contracts to specify permitted uses of AI generated replicas, while AB 1836 restricts commercial replicas of deceased performers without permission from their estates. 

The problem is that celebrity voice rights still vary across the country. The Senate Judiciary Committee announced in June 2026 that it advanced the NO FAKES Act by voice vote. However, the official Congress.gov tracker continued to list the measure as introduced rather than enacted. The proposal is designed to create a federal right against unauthorized digital replicas, but AI voice cloning remains governed largely by a patchwork of existing state laws, contracts and federal claims. 

Not every use will produce a winning lawsuit. The Tennessee legislation expressly recognizes protections for news, criticism, satire, parody and other speech covered by the First Amendment. The U.S. Copyright Office similarly recommended that any federal digital replica law balance personal rights with free expression. A fake voice used to deceive fans into buying something presents a stronger case than an obvious parody that no reasonable listener would mistake for an official performance. 

The Tupac Shakur estate showed how quickly legal pressure can work without a final court ruling. According to Billboard Canada, the estate threatened legal action over AI vocals used on Drake’s “Taylor Made Freestyle” and demanded its removal. Drake removed the track from social media shortly afterward. The incident did not create binding legal precedent, but it demonstrated how publicity rights, estate control and a rapid cease and desist demand can stop distribution before a lawsuit reaches trial.

Black artists have also warned that the fight goes beyond one fake song. In written Senate testimony, FKA twigs said AI tools can have artistic and commercial value when controlled by the artist, but using someone’s identity without consent creates exploitation. Ice Cube called the practice “demonic” and said he would sue anyone who created AI vocals using his voice. SZA told i-D that AI’s impact was happening disproportionately with Black music, while criticizing generated songs that reduce Black expression to stereotypes. 

The strongest AI voice cloning cases involve a clearly identifiable voice, no meaningful consent, commercial distribution and a presentation that suggests the celebrity participated or approved. Artists may rely on publicity rights, false endorsement laws, contracts and copyright claims involving actual recordings or compositions. Celebrities can legally fight back, but until Congress creates a uniform federal remedy, the outcome will depend on the use, the evidence and the state where the battle takes place.