Dario's been openly talking how worried he is about China and labs getting synthetic training data off their models, for years. Most recently in relation to "Mythos level" capabilities.

Not really distillation, just synthetic training data.

What moat?