Instead of using clean data, Adobe leadership followed an “ask forgiveness not approval” approach in using a dataset with pirated literary works to develop its AI services, according to the stockholder derivative complaint filed by a pension fund April 24.
By adopting an unlawful strategy to train its SlimLM model, executives and board members exposed Adobe to potentially massive liability in copyright infringement lawsuits, the SEIU Pension Plans Master Trust told the US District Court for ...