A Proposal for Regulating AI Training Data and Sharing Profits

Summary

This opinion post argues that the current relationship between AI companies, workers, and content creators is economically unsustainable. It identifies two problems: companies may seek to automate a large share of jobs, potentially leaving many consumers without employment, while also using other people’s and companies’ content to generate private profits. The author proposes rebuilding data models around explicit rights and traceability. Model content would be marked, and creators would be paid in proportion to each item’s use in training. Data without payment would be excluded unless the creator grants a free license, and commercial data would require explicit permission. For material with no identifiable owner, such as Shakespeare’s works, profits would instead be paid to the relevant country. The author calls for these rules to be enforced worldwide. The intended result is to keep people affected by automation connected to the AI economy, give open-source developers, photographers, artists, and other creators a share of AI profits, and preserve incentives to continue producing content. The post presents these outcomes as the rationale for the proposal; it does not provide an implementation mechanism, economic analysis, or evidence that the proposed system would produce them.