As TechCrunch AI outlines, the emerging legal framework seems less about the act of ingestion and more about the nature of the output. This suggests a strategic shift for AI developers: training on a corpus to create a directly competing service is the highest-risk activity. The more transformative or distant the end product is from its training data's commercial purpose, the safer it likely is.
This could incentivize AI firms to avoid building tools that directly substitute for the copyrighted works they consume, potentially shaping the entire product roadmap of the industry based on litigation risk rather than capability.