Everyone’s building with LLMs.

Nobody’s talking about, who owns the training data.

Should contributors be rewarded for what models learn from them?