This is largely what Google (et al.) do -- crawling the web and informing models as to the associations between various things.

Copyright appears to mostly pertain to reproduction, which Google and the like steer clear of. But nonetheless they do use content to inform various machine-learning models. Where does the law stand on this? And, even if it were to be legally questionable, how would it be discovered without any trace of reproduction?

An example scenario would be a "Medical Knowledge" AI that crawls and scrapes the entire web looking for information regarding medical ailments. It finds information on Wikipedia, forums, twitter, etc. and starts to create associations between ailments, symptoms and remedies. Any specific piece of content would only have had a very slight effect on the overall weightings within the various ML models, so what claim, if any, would a plaintiff have in laying stake to "their" copyrighted content?



You must log in to answer this question.

Browse other questions tagged .