Comment by michaf
6 hours ago
Is there such a license? Or any license with special clauses for LLMs? Is it enforcable? Could someone 'poison' an LLM training run with injecting just one such licensed document? I am genuinely curious about what levers exist (or are conceivable) to protect your own IP from becoming LLM training data, if regular copyright does not qualify.
This isn't the kind of thing you can do with a license, as long as training a model doesn't require a license. Now, that's an open question legally in the US, and there are active lawsuits, but that does seem like the way it's most likely to play out.