These AI models made a song, and it’s actually a banger

DJ 3LAU is moving into AI music, and he teamed up Grimes AI for one track.
DJ 3LAU is moving into AI music, and he teamed up Grimes AI for one track.
Nick Marfing at Altimira

Happy Friday. It’s tech reporter Alexandra Sternlicht. 

Remember last year when Grimes announced that anyone could use the AI version of her voice so long as she got a portion of the royalties? Well, on Tuesday DJ 3LAU released a song combining the GrimesAI with an artificially intelligent version of his own music—and it’s quite a bop.

But it’s more than just an EDM banger with velvety vocals; it’s an advertisement for curious technologists to tap the 3LAU AI and GrimesAI models to produce their own tracks. Today Justin Blau, aka 3LAU, released his entire library of songs made over 12 years as an open-source AI model so users can see how the DJ who attracts over 1 million monthly Spotify streams makes his tunes, and utilize the technology for their own 3LAU AI-generated songs. 

He’s tracking users and controlling distribution through blockchain technology via his Web3 company Royal, but, unlike Grimes, Blau is not planning to make money from users releasing 3LAU AI-made songs. Right now, it costs just 0.011 ETH—or about $25—for unlimited access to the 3LAU AI library for unlimited use of 3LAU AI. To protect from a flood of 3LAU AI-generated songs: He’s capping the number of licenses at 333, meaning that the 333 people who first buy access to the 3LAU model will have the ability to make unlimited 3LAU AI music. 

“I was like, ‘Screw it. I’m going to open everything up to these engineers so they can figure out how I do what I do,’” Blau told Fortune. 

At this infant stage of AI music, it’s hard to say what this AI model collaboration means for the music industry. Blau believes AI music models like 3LAU AI will first disrupt the electronic and pop music fields as these genres are more “mathematical” than others. 

This will likely raise alarm bells for a sect of the music community that fears a loss of artistry from AI-generated songs. The most notable example of this was the intense backlash from Ghostwriter977’s song “Heart on My Sleeve” which utilized AI-made vocals in the likeness of Drake and The Weeknd. The song was ultimately deleted from streaming platforms as Universal Music Group, which represents the human versions of Drake and The Weeknd, argued it was unlawful to use these vocals without licenses from the artists and labels. 

But for artists like Grimes, who is allowing anyone to license an AI version of her vocals so long as she retains 50% of royalties, AI models collaborating is a step in the right direction. “We are already cyborgs,” says Grimes’s manager Daouda Leonard, a music industry veteran who has managed the careers of DJ Snake and Grimes, among others. “To have machines that are going to be on your side or part of you, enhancing what you’re already capable of, I think that’s a beautiful thing.”

It stands to reason that AI’s dominance over the music industry will be slow and only partial—for now. Blau notes that live performances and relatable human vocalists will always be in demand. “To be honest, I think we’re way too early,” says Blau. “It’s gonna be a really long time before this whole concept goes mainstream.”

That said, Blau says he sold all 333 licenses for 3LAU AI five minutes after launching the product.

Here’s what else is going on in tech news today.

Alexandra Sternlicht

Want to send thoughts or suggestions to Data Sheet? Drop a line here.

NEWSWORTHY

Palantir in Israel. Palantir, the big data analysis and AI company cofounded by Peter Thiel, is partnering with Israel to supply the country with tech to aid its war effort, Bloomberg reported. The agreement follows a visit from Thiel and CEO Alex Karp to Israel last week. 

Testing AI. The U.K. government will hold a series of tests to determine if AI companies uphold their commitments to prevent harm and if its new AI Safety Institute is effective in identifying risks, the Financial Times reported. The outcome of the tests will determine if the government passes AI-focused legislation. The U.K. previously said it will avoid regulating AI in the short term over fears it could stunt industry growth.

Demand for corporate ChatGPT. OpenAI has secured 260 business customers for its corporate chatbot product ChatGPT Enterprise in the first four months after its launch, Bloomberg reported. The companies have a combined total of 150,000 employees that have signed up to use the product.

IN OUR FEED

“We were disappointed with the New York Times. We actually thought there was an opportunity there for what could have really been an important partnership. We're not going to be distracted from the companies that want to work with us.”

OpenAI chief operating officer Brad Lightcap said in an interview with Bloomberg about how the company intends to continue making licensing deals with publishers in the face of a major lawsuit by the New York Times, which accused OpenAI of copyright infringement. Lightcap added that 99% of OpenAI’s engagements with media companies have been positive and collaborative.

IN CASE YOU MISSED IT

Exclusive: Slack CTO Cal Henderson is out and will be replaced by Salesforce cofounder Parker Harris, by Kylie Robison

A rare Netflix skeptic on Wall Street says it fails the grandmother test: The winner of the streaming wars will have to do too much to stay on top, by Paolo Confino 

Apple undergoes its biggest board shakeup in years as 2 longtime directors depart, including Al Gore, by Mark Gurman and Bloomberg 

Garry Tan says founders ‘have to be in San Francisco’ as Y Combinator ditches Mountain View headquarters for the big city, by Jessica Mathews

After AI-generated ‘George Carlin’ routine, late comedian’s daughter warns other entertainers: ‘They’re coming for you next’, by Steve Mollman

The $40 billion AI mania sprouts a tech incubator program in the unlikeliest place: Major League Soccer, by Paolo Confino 

Laid-off Big Tech workers are back on the job market. These hard truths I’ve come to learn can help them land jobs in startups, by Alex Atzberger

BEFORE YOU GO

Questioning connected cars. The Federal Communications Commission is concerned that connected car apps allow abusive partners to track and harass their victims, the New York Times reported. On Thursday, the government body sent letters to nine of the biggest car companies—including General Motors, Toyota, and Tesla—requesting information about how they’re addressing this problem. It also contacted phone carriers Verizon, AT&T, and T-Mobile about their role in supporting connected car apps. 

These apps allow a vehicle owner to remotely adjust the temperature, honk, view the car’s location, and more, but a Times investigation last month determined they also help abusive partners stalk their victims.

This is the web version of Data Sheet, a daily newsletter on the business of tech. Sign up to get it delivered free to your inbox.