Law of Capture applied to gen AI intellectual property

This is a pretty interesting paper that I found only because they referenced my AI-assisted books in it. It’s by Arjun Padmanabhan and Tanner Wadsworth and explores two ideas from the history of property law, the Law of Capture and the Doctrine of Accession. Both are interesting, but this part about the Law of Capture is of the greatest interest to me when applied to the realm of AI-assisted creative works.

The law of capture provides that a person can acquire title to otherwise un-ownable property by reducing it to his dominion and control. It is a necessary outgrowth from the law of first possession; necessary because “possessing” a piece of land does not necessarily mean possessing all the things on or under it. Wild animals, fish, and birds are all “fugitive resources” that travel from place to place without respect for human-created property lines. These things are presumably not owned by anyone until they are captured, at which point they belong to the captor.

In the classic example, a hunter can own a wild fox, but only once he has removed it from its natural state by killing or trapping it. Similarly, nobody owns a whale until it has been fatally harpooned. Once a fisherman has harpooned the whale, however, that fisherman owns it, no matter who discovers the carcass or where it washes up on shore.

This approach to owning the “fruits” of explorations in latent space makes total sense to me… The comparison of “foraging” in latent space to hunting & trapping in higher-dimensional space feels intuitively right to me.