Comment by mariojv

To me, it seems like more of a competitive issue for OpenAI if part of their secret is the ability to synthesize good training data, or if they're purchasing training data from some proprietary source.