Sony Music v. Suno Heads to a Summary-Judgment Showdown That Could Redefine AI Fair Use

--- headline: "Sony Music v. Suno Heads to a Summary-Judgment Showdown That Could Redefine AI Fair Use" slug: "sony-suno-summary-judgment-copyright" category: "policy" story_number: 13 date: "2026-07-23" ---

Sony Music v. Suno Heads to a Summary-Judgment Showdown That Could Redefine AI Fair Use

A federal courtroom in Boston has quietly become the most important address in the artificial-intelligence economy. There, in the U.S. District Court for the District of Massachusetts, Chief Judge F. Dennis Saylor IV is presiding over Sony Music's copyright case against Suno, the AI music startup that has raised more than $400 million this year at a $5.4 billion valuation. Both sides are expected to move for summary judgment on a single question that reaches far beyond music: whether training an AI model on copyrighted works without a license is fair use.

The answer will not stay in the music aisle. The same legal test governs the models generating text, images, code and video, which is why every AI developer and every rightsholder is now watching a docket about pop recordings.

What the fingerprints found

The case began in June 2024, when the Recording Industry Association of America sued Suno and its rival Udio on behalf of all three major labels, accusing them of "mass infringement." Suno's original complaint listed 560 works. Then the labels went looking for the rest.

Because Suno declined to disclose exactly what it had ingested, the labels' experts were given supervised access to a secure room at the company's outside counsel and spent two weeks creating digital fingerprints of the audio in Suno's training data. Using Audible Magic's content-recognition technology, which identifies recordings by their pitch distribution, harmonic content and rhythmic patterns, they matched those fingerprints against a reference database.

The result reframed the litigation. On May 21, Universal Music Group and Sony asked the court to expand the complaint to 61,026 specifically identified recordings, a figure they described as "only a small fraction" of the total matches. Suno has largely conceded the underlying facts, admitting in its answer that building the service "required showing the program tens of millions of instances of different kinds of recordings," which "presumably included recordings whose rights are owned by the Plaintiffs."

That concession is strategic. By stipulating that its training set was built substantially on copyrighted material, Suno has bet its entire defense on the law, not the facts. Its argument is that training is "transformative" because the model learns patterns to generate new songs rather than reproducing the originals.

The dollars and the delay

The size of the list is not a detail; it is the ballgame. U.S. copyright law caps statutory damages at $150,000 per work for willful infringement. Under the original 560-work complaint, Suno's theoretical maximum exposure sat around $84 million. Attach 61,026 recordings and that ceiling rockets above $9 billion.

Suno is fighting hard to keep the number small, and to keep the schedule moving. In a June 4 filing, its lawyers accused the labels of running "a too-familiar page from the standard playbook of aggregate music rightsholders: file an action asserting 'representative' works, let the litigation proceed through discovery for years, then attempt to expand the list of works exponentially at the close of fact discovery." After "two years of extensive fact discovery," the filing argued, "Suno is entitled to a timely consideration of its fair use defense."

The labels counter that Suno created its own problem. The "cause of any delay," they wrote, "was Suno's ongoing refusal to provide Plaintiffs with the data in its possession."

Suno has a fresh precedent to lean on. In the parallel Udio case in the Southern District of New York, Judge Alvin K. Hellerstein on June 29 refused to let Sony add more than 30,000 recordings, keeping that suit at 333 works. "Adding more than 30,000 works near the close of document discovery would require substantial additional production and review, generate further disputes, and materially alter the scope of the case before me," Hellerstein wrote. Days later, Suno counsel Britt Lovejoy of Latham & Watkins put that ruling in front of Judge Saylor, urging him to "deny Plaintiffs' motion here as well." Saylor has not yet ruled.

Why the precedent travels

The fair-use fight will be decided against a backdrop of recent AI wins for developers. In Bartz v. Anthropic, Judge William Alsup called the use of books to train a model "spectacularly" transformative, and Kadrey v. Meta reached a similar conclusion. But those were text cases. Anthropic still paid a reported $1.5 billion to settle claims tied to pirated books, a reminder that how the training data was acquired can matter as much as what the model does with it.

The labels argue the music context is different in a way that could break Suno's analogy: unlike a chatbot's prose, Suno's output competes directly in the same market as the recordings it trained on. That market-harm distinction is the fourth fair-use factor, and it is where a music ruling could diverge sharply from the book cases.

The stakes are systemic. Legal trackers now count well over 100 active AI copyright suits in the United States, with claimed exposure running into the tens of billions of dollars. A clean ruling from Judge Saylor on whether unlicensed training is fair use would be the first hard signal for that entire docket.

What to watch

The near-term flashpoint is Saylor's decision on the 61,026-recording expansion, which sets both the damages ceiling and the pace toward summary judgment. Fact discovery and depositions close September 30, 2026, with dispositive motions due April 9, 2027, so the fair-use question will be fully briefed rather than snap-decided.

There is also a settlement current running beneath the litigation. Warner Music settled with Suno in November 2025 and struck a licensing partnership; UMG settled with Udio. A fair-use ruling would reset the leverage in every one of those negotiations, and abroad Suno still faces claims from Germany's GEMA and Denmark's Koda. Whichever way Saylor rules, the industry's next contracts will be written in its language.