In the age of the "fair use" debate around training models using copyrighted content, how will authors of said content receive credit, attribution, and money for their content that is used to generate new content. As the entire "content supply chain" will increasingly be LLM-based, instead of web-page search engine based, what are the most promising techniques to "track and attribute" whose content was used to generate what content?
Has blockchain provenance finally found a use case? Invoking certain FFN's (and tracing which ones were invoked) in some derivative of Mixture of Experts?