New bill mandates AI foundational models to disclose training data sources: Report

Legislators in the U.S. have introduced a bill that would require creators of foundational AI models to disclose the sources of their training data. The bill, called the AI Foundational Model Transparency Act, aims to provide transparency to copyright holders by allowing them to determine if their data was used in the creation of AI models. The Federal Trade Commission would collaborate with the National Institute of Standards and Technology (NIST) to establish reporting rules for training data transparency. Companies that develop foundational AI models would be obligated to disclose the data sources they used and provide information on their efforts to prevent the models from disseminating inaccurate or harmful information. This bill aligns with the AI executive order issued by the Biden administration, which aims to establish standards for AI safety, security, privacy, and civil rights. In recent months, several lawsuits have been filed against companies involved in AI model development, alleging copyright infringement. One such case involved Getty Images suing Stability AI for allegedly copying and processing millions of copyrighted images without permission.