Australia and Japan Diverge on How to Shield Creators from Generative AI

Two close US allies, Australia and Japan, are moving in opposite directions on how to protect creative intellectual property from generative AI systems. According to recent policy reporting, Australia is working toward an “exclusion first” framework that would let creators opt their works out of being used as training data for AI models under defined conditions. In practice, that would strengthen artists’ and writers’ ability to say no when their books, images or music are scraped into datasets for large models.

Japan, by contrast, is leaning toward a “transparency first” strategy. Rather than creating broad opt‑out rights, cultural and technology agencies are exploring rules that would force AI developers to disclose more about what datasets they use and under what terms, and to clarify how copyrighted material is handled. The split reflects the same frustration in both countries: today’s generative models are trained on vast troves of creative work, yet rightsholders have almost no visibility into how their output was ingested or monetized. Australia’s approach could give creators more direct control but might also restrict access to data for researchers and smaller AI firms. Japan’s path aims to keep innovation flowing while relying on disclosure and accountability – but leaves open hard questions about compensation and revenue‑sharing. Together, they highlight an emerging fault line in global AI copyright governance.