Sampling has become increasingly complex and expensive, with the need for proper licensing and rights clearance. The discussion highlights the distinction between input and output in copyright issues, particularly in relation to AI-generated products, emphasizing that the real concern lies in how models are trained on existing sound recordings rather than just limiting prompts.
