Welcome to Eye on AI, with AI reporter Sharon Goldman. In today’s issue: The small business owners managing whole armies of AI employees…Meta keeps delaying the release of its new AI Model to developers…How courts are coping with a flood of AI-generated lawsuits.
Yesterday, AI music generator Suno announced a new $400 million funding round at a $5.4 billion valuation, the latest sign that investors believe AI-generated music is here to stay. Ironically, just a few days earlier, I was sitting in a 14th-century French château in the Loire Valley with a dozen other songwriters, geeking out over lyrics, melodies, and rhyme schemes at a bucket-list retreat.
The contrast, of course, is striking: On one side, a startup built on the idea that anyone can create a song in seconds. On the other, people traveling across the ocean to spend a week doing it the slow way.
But after spending the week far away from Silicon Valley, both physically and metaphorically, I’m not convinced either vision settles the bigger questions about AI music.
If you’ve ever tried Suno (and I strongly suggest you do), it’s pretty awesome. With a simple text prompt, it can generate remarkably convincing songs in seconds. The possibilities can be breathtaking, hysterically funny, and viral, like the trend this spring turning messaging threads with friends into songs, and then sharing them on TikTok. Personally, I enjoyed this Suno-generated song that turned text messages between Sam Altman and Mira Murati from the day Altman was fired from OpenAI in 2023—messages later introduced as evidence in Elon Musk’s lawsuit against OpenAI—into a song in the style of the Broadway musical Hamilton.
In a blog post announcing the new funding, Suno wrote that its initial focus was simple—to allow more people to experience the joy of making music. “In recent months, we’ve seen Suno become part of culture in ways that continue to surprise us,” said the post. “Family members are turning text threads, group chats, and inside jokes into songs. People are writing songs for birthdays, graduations, and even work events. Viral trends helped propel Suno to #1 in the App Store’s Music category in dozens of countries.”
It cited meaningful use cases such as patients in hospice care using Suno to leave songs behind for loved ones; therapists helping teens navigate mental health challenges through music creation; caregivers for people with dementia and Alzheimer’s creating personalized songs tied to memories and familiar voices.
But while the technology is impressive, and clearly has an audience, what’s less clear is whether that audience is large enough to support the kind of venture-scale business implied by a $5.4 billion valuation. Could Sumo become a daily habit like Spotify, Netflix, or ChatGPT?
Demand isn’t the only uncertainty: Sumo and its competitors are trained on vast numbers of human-created songs, and the legal battles over whether that training is lawful are far from resolved.
I’ve been covering those disputes since Suno and rival Udio emerged in 2023. As I wrote in a 2024 Fortune essay, Sony Music Group—which represents artists including Adele and Beyoncé—warned hundreds of AI companies not to train models on its content without permission. Artists including Billie Eilish, Nicki Minaj, and Stevie Wonder signed an open letter arguing that “this assault on human creativity must be stopped.”
Suno and Udio have both acknowledged using copyrighted recordings to train their models, but argue that the practice is protected under the legal doctrine of fair use. Copyright holders disagree. Universal Music Group, Sony Music, and Germany’s GEMA have continued to pursue legal action against Suno, while [hotlink]Warner Music Group[/hotlink] reached a licensing agreement with the company last year.
The scale of the dispute has only grown. When the record labels first sued Suno in 2024, they alleged the company had trained on roughly 560 copyrighted works. Last month, they sought to amend their complaint to claim that more than 61,000 additional songs were used without permission. Meanwhile, both Suno and Udio have asked courts to keep the size of their training datasets confidential, arguing that the information could help competitors build rival products.
For now, investors appear willing to look past those legal uncertainties. Suno remains one of the most popular music apps in the world, and according to fundraising materials obtained by Billboard, users were generating more than 7 million songs per day at the time of its latest financing.
Menlo Ventures, which led Suno’s Series C last fall, said in a blog post that it was “thrilled” to double down on Suno in its latest round, and pointed out that “Every major consumer platform is built on a new behavior. TikTok made short-form video consumption mainstream. Netflix changed how we watch TV. Suno is doing something different: making creation itself a form of entertainment.”
Suno appears to be banking on future partnerships with the music industry, including a new music model. “We believe there’s a huge opportunity to create new experiences for fans while helping artists reach audiences, build community, and unlock new creative and economic possibilities,” the company’s blog post read.
Whether or not Suno succeeds, I suspect music will continue to exist on a spectrum of technology use, which has existed since recorded music began with the invention of the phonograph. Even among my cohort at the château, there were songwriters who routinely relied on online beats, digital recording software, pitch correction, social media distribution, and countless other technologies. Musicians have always adopted new tools when they found them useful.
As The Verge reported a few months ago, even some Nashville songwriters are already experimenting with AI tools to generate demos and explore ideas more quickly. I suspect we’ll see more of that.
But will making music become a mass-market form of entertainment in its own right? I prompted Suno to write a yearning country song about that—and this entire essay. The result is both weird and impressive, but also doesn’t answer the question. Only time will do that.
With that, here’s more AI news.
Sharon Goldman
sharon.goldman@fortune.com
@sharongoldman
A new AI-powered computer worm could prove to be the stuff of cybersecurity nightmares -- by Sharon Goldman
Exclusive: Nvidia snaps up Kumo AI in latest acquisition -- by Sharon Goldman
Cognizant CEO is swimming against the tide on AI: he’s hiring over 20,000 graduates this year and says AI tokenmaxxing is a ‘vanity metric’ -- by Preston Fore
Nscale has raised billions to power Europe’s AI ambitions. Now the startup must prove the hype can survive reality -- by Beatrice Nolan
Exclusive: Apoha, a startup building AI models for creating new materials, emerges from stealth with $36 million in funding -- by Jeremy Kahn
The small business owners managing whole armies of AI employees. This New York Times Magazine piece is a fascinating look at how small businesses are using the open-source agent platform OpenClaw, which allows AI systems to access files, log into software, communicate with customers, conduct research, manage schedules, draft legal documents, process orders, and handle other office tasks with minimal supervision. From law offices to apartment managers, these early adopters report significant productivity gains and even reductions in staffing needs, but the technology remains unreliable, expensive, and vulnerable to security risks.
Meta keeps delaying the release of its new AI Model to developers. According to the Wall Street Journal, Meta has repeatedly delayed the release of the API for its newest AI model, Muse Spark, pushing back a launch that was initially expected shortly after the model debuted in April. The setback highlights a growing challenge for Meta as it pours as much as $145 billion into AI infrastructure and seeks ways to monetize its frontier models. Unlike Meta's previous open-source releases, Muse Spark is a closed model, meaning developers can only access it through an API—the same approach that has helped drive revenue growth at OpenAI and Anthropic. While Meta says it plans to release the API this month, the delays underscore the difficulty of translating AI research breakthroughs into commercial products, even for one of the world's largest technology companies.
How courts are coping with a flood of AI-generated lawsuits. The MIT Technology Review highlighted a new study that suggests AI is reshaping access to the U.S. legal system, with self-represented litigants increasingly using chatbots to draft court filings, contributing to a sharp rise in lawsuits filed without lawyers. Researchers found the share of federal civil cases brought by people representing themselves rose from 11% in 2022 to 16.8% in 2025, while judges report that AI-assisted filings are often clearer and easier to understand—even if they still contain hallucinations and factual errors. Yet the technology isn't improving litigants' odds of winning, and it is raising a host of new legal questions, from whether conversations with ChatGPT deserve attorney-client-style protections to who should be liable when AI systems provide bad legal advice. But judges say AI is already making the legal system more accessible to people who otherwise might not have been able to navigate it at all.
That's how much Google parent Alphabet increased its planned equity raise, up from the $80 billion announced just two days earlier, as it seeks additional capital to fund its escalating AI infrastructure investments. The offering, which includes a $40 billion at-the-market share sale program and a $10 billion investment from Berkshire Hathaway, is set to become the largest equity capital markets transaction in history, surpassing Petrobras's roughly $70 billion stock sale in 2010. The move underscores the enormous costs of the AI race, as Alphabet ramps up spending on data centers and its homegrown TPU chips to compete with Nvidia and meet growing demand for AI computing power.
June 8-10: Fortune Brainstorm Tech, Aspen, Colo. Apply to attend here.
June 17-20: VivaTech, Paris.
July 6-11: International Conference on Machine Learning (ICML), Seoul, South Korea.
July 7-10: AI for Good Summit, Geneva, Switzerland.
Aug. 4-6: Ai4 2026, Las Vegas.
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。