UMG and Sony can pursue Suno over alleged YouTube ‘stream ripping’ used to train AI

# What happened

# What the labels allege

# What the judge ruled Judge Saylor did not determine guilt. He ruled the labels' amended complaint presents a plausible claim that warrants further proceedings and evidence-gathering. The court said it will require more factual development to decide how the YouTube technology and the alleged downloading tools actually operate. That factual phase—discovery—is scheduled to close on September 30, 2026.

The stream-ripping claim is about how files were obtained, not whether training an AI on copyrighted recordings is legally permissible. UMG and Sony already brought separate copyright claims against Suno alleging unauthorized use of recordings for model training. Whether such training can be fair use remains an open question the court will address later, likely through summary judgment after discovery.

# Why independent artists should pay attention Independent musicians who upload music to platforms such as YouTube could be affected. The labels' theory draws a line between public availability and permission to copy recordings into a commercial AI training set. Independent artists often rely on public platforms for exposure, but public hosting does not necessarily equal consent for reuse in training commercial AI.

# Related and parallel legal action

# Suno's response and company moves Suno has contested the legal theory behind the DMCA claim, arguing that the statute addresses access controls rather than copying, and disputing whether YouTube's technical measures meet the DMCA's definition of an access control. Separately, Suno has begun making licensing moves: it announced a deal with BMG described as opt-in for artists and said it will retire earlier models and introduce new industry-partnered models. Those licensing steps coexist with ongoing litigation.

# What's next Fact discovery runs through September 30, 2026. After that, the parties are expected to seek summary judgment on whether training models on copyrighted recordings without licenses can qualify as fair use. The stream-ripping claim will be resolved through the same factual record, once the court evaluates technical evidence about YouTube's systems and any tools Suno used.

# Practical takeaway for creators Uploading music publicly does not automatically grant permission for use in commercial AI training. Rights holders are actively testing whether technical circumvention claims and copyright law can limit how AI companies gather and use recordings. Independent creators should monitor developments and consider licensing terms on platforms and direct agreements if they want to control downstream uses.