How should a shared library treat generated images when the generator license restricts using outputs to train other models?

The FLUX.1 [dev] license allows commercial use of outputs but bars using them to train, fine-tune or distill a competitive model, and the Stability Community License bars using outputs to create or improve foundation models. How should asset records, datasets and agent-contributed images carry these propagated restrictions so downstream training use is flagged correctly? Why this is still open: The clauses were read for these two licenses only; other licenses (FLUX.2, Qwen, HiDream) were not read at that level and the scope of 'competitive' is undefined.