Kendrick Lamar AI: The Fight Over a Copied Voice

Kendrick Lamar AI is not an album, a product or a side project from the rapper himself. The term covers songs and short clips made with artificial intelligence that copy Kendrick Lamar’s voice, his flow and his delivery, then circulate online as if he recorded them. He did not. No label released them, no studio booked the session, and in most cases the first time his team heard a track was the same time fans did.

The story matters because his voice became a test case for a much bigger question. If software can rebuild a famous voice from old recordings, who controls that voice: the artist, the fans who build the copies, or nobody at all? Across three years, fake Kendrick tracks, a diss record built from two other copied voices, and a wave of new state laws gave that question real answers. This article walks through what actually happened, what the law says today, and how fans can tell a fake from the real thing.

The demo that started the argument

In February 2023, a short clip began moving through hip hop social media. A man built a song from nothing on camera, then ran his own voice through a filter that made him sound like Kendrick Lamar. The result was rough around the edges, but the likeness was close enough to unsettle people who make music for a living.

One of them was Young Guru, the Grammy winning engineer behind many of Jay-Z’s records. Sharing the clip, in comments reported by Okayplayer, he wrote that the world was in a very groundbreaking but dangerous moment, and added the line that still frames the whole debate: “It’s not the tech, it’s the evil that men do with the tech.”

His sharper point was legal. “At this present moment you can’t copyright a voice,” he wrote. “You can copyright a song, or a speech but not the voice itself.” Copyright, the system that protects songs, recordings and lyrics, had no box for the sound of a person. “The law has to change to protect everyone not just public figures,” he added. One clip did more than any other to start the Kendrick Lamar AI argument, because it named the gap the next three years would fight over.

The nuance fans missed is that Kendrick had already played with this family of technology himself. His 2022 video for The Heart Part 5 used face swapping, built with the studio Deep Voodoo, to turn him into figures such as Kanye West and Will Smith as he rapped. The difference was consent. That was his video, his face and his choice. The AI tracks that copy his voice have none of the three.

How a voice clone gets made

The tools behind ai voice cloning rappers are not magic, and the method explains both the quality of the fakes and their limits. A cloner starts by gathering recordings: album vocals, interviews, freestyles and live clips. Famous rappers leave hours of clean audio in public, which is why stars get cloned first and unknown singers almost never do.

Software then studies that audio. At its heart sits a neural network, a program that learns patterns from examples instead of following fixed rules. It breaks the voice into traits: pitch, rhythm, breath, the way a word bends at the end of a line. What it stores is not a recording. It is a recipe for rebuilding the voice on demand, able to say things the singer never said.

From there, the cloner performs the song. Some sing or rap it themselves and let a filter repaint the performance in the target voice. Others type the lyrics and let text tools, cousins of natural language processing, turn written words into a sung performance. The output can be startling, yet close listening usually betrays it: breaths land in odd places, emotion stays flat, and dense, shifting flows blur. The voice is copied. The judgment behind the real performances is not.

The spring the fakes reached the charts

That spring, the Kendrick Lamar AI story stopped being only about Kendrick. The demos had been assembled with the same machine learning algorithms behind his fakes, and in April 2023 the trick reached the top of pop. A TikTok creator who called himself Ghostwriter uploaded Heart on My Sleeve, a song built from AI voices of Drake and The Weeknd. It sounded like a lost collaboration and spread like one: roughly 600,000 plays on Spotify, about 275,000 views on YouTube and millions more on TikTok, as Billboard reported.

Universal Music Group, the label behind both singers, moved fast. Within about two weeks the song was gone from Spotify, Apple Music, YouTube and the other major services. UMG said training AI on its artists’ music broke its agreements and copyright law, and asked which side of history the industry wanted to be on: “the side of artists, fans and human creative expression, or on the side of deep fakes, fraud and denying artists their due compensation.”

The takedown taught the industry its first real lesson. The services had no clear rule against a fake voice as such. They acted because the label framed the song as a rights violation and asked them to. Removal, in other words, ran on complaints, not on an automatic ban. That pattern, complain first and remove second, still shapes how fake tracks are handled today.

Taylor Made: the clone enters the feud

A year later, the Kendrick Lamar AI fight moved from fan experiments to the center of the biggest rap feud of the decade. Drake and Kendrick Lamar were trading diss records in the spring of 2024. On April 19, Drake posted Taylor Made Freestyle to Instagram, a track in which AI generated voices of Tupac Shakur and Snoop Dogg urged Kendrick to answer him. Drake’s own verse closed the song.

The choice of voices was the provocation. Tupac is a hero in Kendrick’s writing, and Snoop is a West Coast elder. Using their voices to pressure him crossed a line the fan demos never had: it put words in the mouth of a dead artist who could not object.

Tupac’s estate answered on April 24, 2024. Its lawyer, Howard King, sent Drake a cease and desist letter, which Billboard published. “The Estate is deeply dismayed and disappointed by your unauthorized use of Tupac’s voice and personality,” it read. The letter called the record “a flagrant violation of Tupac’s publicity and the estate’s legal rights” and “a blatant abuse of the legacy of one of the greatest hip-hop artists of all time.” The estate said it would never have approved the use, and noted that aiming Tupac’s voice at Kendrick, a friend of the estate who had always shown him respect, “compounds the insult.” Drake had 24 hours to take the song down. By April 26, the post was gone.

Snoop Dogg, alive and able to speak for himself, reacted with a short, confused video: “They did what? When? How? Are you sure?” Kendrick later noted that Snoop had shared the track, a grievance he aired on wacced out murals, the opening song of his album GNX. The diss itself is now a footnote. Its real legacy is legal: it showed that an estate letter, not a court case, is still the fastest weapon against a voice clone.

Real release or fake? How fans check

For listeners, the practical question is simpler: how do you know a new Kendrick track is real? A fake kendrick lamar ai track usually fails the paperwork test long before it fails the listening test. Real releases arrive with a label announcement, full credits and a place on his official artist pages. Fakes surface first as clips on fan or anonymous accounts, with no credits, no label and no comment from his team.

Sound still offers clues. Cloned voices handle simple lines well and struggle with the dense, shifting flows that made him famous. Breaths sit in the wrong places, backing vocals smear, and the emotional swerves of a real performance flatten out. When a track feels close but not quite right, it usually is not right. Checking the artist’s official page takes under a minute and settles most arguments.

What the law does today

Young Guru’s complaint in 2023 was that no law owned this problem. The map is clearer now, if still patchy. Copyright protects recordings and songs, not the sound of a voice. The right that covers a voice in the United States is the right of publicity, a state level rule that stops others from using a person’s identity for gain. Around 30 states have publicity laws on the books, and they differ from state to state. It is the gap every Kendrick Lamar AI case since 2023 has run into.

Tennessee moved first and furthest. Its ELVIS Act, signed in March 2024 and in force since July 1, 2024, was the first US law to name a person’s voice, including a simulated voice, as a protected property right. It also reaches the tools themselves: software whose main job is producing a specific person’s voice or likeness without permission. California followed with AB 1836, in force since January 2025, which requires estate consent before a digital copy of a deceased performer’s voice is used in a recording. The Tupac case sits squarely in that territory.

A federal answer is still missing. The NO FAKES Act, tracked on Congress.gov, would give every American a right over digital copies of their voice and face, and would make platforms remove fakes after a valid notice. It was reintroduced in April 2025 and cleared the Senate Judiciary Committee in June 2026, but as of this writing neither chamber of Congress has passed it. Until that changes, an artist’s protection depends on the state they call home and on how fast their label complains.

What the platforms do

Between the state laws sits the layer fans actually meet: platform rules. After the 2023 takedowns, YouTube built a request system that lets its music partners ask for the removal of AI content that mimics an artist’s unique singing or rapping voice. Spotify went on the record in September 2025: vocal impersonation is allowed only when the impersonated artist has authorized it. Smaller services mostly lean on the same notice and takedown pattern that removed Heart on My Sleeve.

The gap in this system is speed and scale. A famous artist with a major label gets a fast response. A session singer or a small independent artist, cloned by a stranger, faces forms, waiting and proof of identity. Voice cloning is also only one front in a wider wave of consumer AI generators, from chatbots to media tools like vheer ai. That gap is the strongest argument supporters make for the NO FAKES Act: one federal rule and one clear process, for everyone.

The October twist

In October 2026, the Kendrick Lamar AI story reached campaign politics. The re-election campaign of Senator Marsha Blackburn of Tennessee shared a 12 second AI generated video that restaged Kendrick’s Super Bowl halftime performance of Not Like Us, with the senator in his place and Tennessee Republicans as her dancers. The clip was captioned “Tomorrow. Ready to roll.”

The backlash wrote itself. Blackburn is a sponsor of the NO FAKES Act and a defender of her state’s ELVIS Act, the two measures built to stop this kind of copy. Her Democratic opponent called her a hypocrite, and the music press picked the story up within hours. Blackburn said a county commissioner had made the clip and called posting it a funsie. Billboard noted that Kendrick’s representatives had not commented. One short video proved both sides of the argument at once: the tools are now trivial to use, and the rules about consent still depend on who is asking.

Conclusion

Kendrick Lamar AI began as a fan trick, a filter laid over a homemade song. It has since forced a diss track off the internet, pushed Tennessee and California to write new law, and put a federal bill in front of Congress. The technology will keep improving, and the copies will keep coming. But the principle that emerged from three years of fights is simple, and in a growing number of states it is now written down: a voice belongs to the person it came from. Borrowing it needs permission, and the age of taking it for free is ending.