Hypothetically, if this network would manage to mirror the Fediverse and then provide a different robots.txt/restrictions than in the original sources to be ingested by OpenAI, would OpenAI be liable?