Stability AI Shifts Audio Model to Fully Licensed Training Data

Stability AI Shifts Audio Model to Fully Licensed Training Data

Stability AI released Stable Audio 3.0 on May 20, trained exclusively on licensed content rather than web-scraped material. The move addresses copyright challenges that have dogged generative AI companies. Whether quality matches web-trained competitors will determine if this becomes industry standard or remains a legal-defensibility play with performance tradeoffs.

Published

Read at another depth