Judge F. Dennis Saylor IV granted permission on August 18 for the labels to pursue a claim under Section 1201(a) of the Digital Millennium Copyright Act. The ruling expands the legal theory in the case, but it does not decide whether Suno is liable, whether the alleged conduct occurred, or whether the company ultimately violated the law.

What the labels allege

In the amended complaint, the labels allege that Suno used tools including YT-DL and YT-DLP to bypass YouTube’s technical protections and download copyrighted sound recordings for training its artificial-intelligence music models.

The complaint describes YouTube’s rolling-cipher protection as an encryption measure intended to prevent external services from directly downloading protected media files. The labels allege that Suno used stream-ripping tools to circumvent that protection and obtain recordings for its training data. Those claims come from the labels’ court filing and remain disputed allegations, not established facts.

The new theory differs from an ordinary copyright-infringement claim. Instead of focusing only on whether copyrighted music was copied or used unlawfully, the DMCA claim examines whether a technological measure controlling access to copyrighted works was bypassed.

What the judge actually decided

The court found that the labels’ allegations were plausible enough to support an amended pleading. At this stage, that means the claim may proceed so the parties can develop a factual record. It is not a finding that the alleged stream ripping occurred, that Suno used the named tools, that YouTube’s protections were circumvented or that Suno infringed copyright.

The judge said the ultimate question would require more evidence about how YouTube’s rolling cipher and the alleged circumvention tools operate. The case therefore remains open on the central factual and legal questions, including whether the tools bypassed an access-control measure covered by Section 1201(a).

Why 61,026 additional recordings were excluded

Judge Saylor separately rejected the labels’ request to add 61,026 recordings to the 560 works already at issue. The court cited the added complexity and delay that such an expansion would create, keeping the current lawsuit narrower.

The ruling leaves open the possibility that the additional recordings could be pursued in a separate lawsuit. That distinction matters because the court allowed the new legal theory to be added while declining to expand the case dramatically at the same time.

Why the case matters for AI music

The immediate change is procedural, but the dispute could influence negotiations among record labels, AI music companies, artists and online platforms. It puts the method of obtaining training material alongside the larger question of whether using copyrighted recordings to train AI models requires permission and compensation.

The case also arrives as the global recording industry develops rules for AI-assisted and AI-generated music. The International Federation of the Phonographic Industry says recordings made with generative-AI services should use properly authorized and lawful tools, remain substantially human-made, comply with applicable rights laws and be appropriately signaled to consumers. The Massachusetts ruling itself, however, does not automatically change the law in other countries.

Fact discovery in the labels’ case is scheduled to close on September 30, 2026, according to the reporting on Judge Saylor’s orders. The parties are expected to continue litigating whether training an AI model on copyrighted recordings without a license is fair use, while separately addressing the new anti-circumvention allegations. Those are distinct issues, and neither has been finally resolved.

Sources

• First Amended Complaint, U.S. District Court for the District of Massachusetts
• Bloomberg Law
• IFPI global AI-chart principles