AI news ·
UChicago tool lets streaming listeners identify AI-generated music in real time
University of Chicago researchers released Quicksilver, a browser extension that flags AI-generated songs in real time while users stream music. AI tracks now make up nearly half of all newly uploaded songs on streaming platforms.
UChicago Tool Identifies AI-Generated Music in Real Time
Researchers at the University of Chicago have released Quicksilver, a browser extension that detects artificially generated songs while users stream music on their computers.
The tool addresses a practical problem: AI-generated tracks now account for nearly half of all newly uploaded songs on streaming platforms, yet most go unheard. Professional musicians perform only slightly better than random chance when distinguishing machine-made music from human compositions, according to UChicago's research.
How It Works
Quicksilver runs as a background process while users listen to music, flagging tracks created with AI generation tools. The extension provides immediate feedback without requiring manual input from listeners.
Stanley Wu, a UChicago graduate student who led development, said the tool addresses a gap in user awareness. "Given that there is so much AI music out there, and that normal users can't tell the difference, giving users a tool to identify these AI songs was a very natural solution," Wu said in a statement.
Who Built It
The UChicago SAND Lab - Security, Algorithms, Networking and Data - developed Quicksilver alongside the nonprofit Ethical Technology and Computing for Humanity (ETCH). ETCH launched this year as a collaboration between UChicago computer science professors Ben Zhao and Heather Zheng.
The SAND Lab focuses on generative AI and LLM detection, security, machine learning, and data mining applications.
The Broader Context
Quicksilver follows two earlier tools from the same team: Glaze and Nightshade, which prevent unauthorized AI training on artwork and creative content.
Wu emphasized the scale of the problem. "At this scale, I think there needs to be more protections in place so that this wave of spammy AI music does not negatively impact human artists," he said.
For developers and IT professionals, the extension demonstrates how detection systems can be integrated into user-facing applications. The architecture offers a model for implementing AI detection in production environments.