Fair Usage in Search and AI

We’ve just published a piece on crawling, scraping, search and AI. The analysis results will be rapidly out-of-date but as of now, are only two weeks old, being based on data from two sources which are themselves constantly being updated.

Does training and deploying AI models on copyrighted works constitute fair use?
We don’t answer our own question, and neither should we. But we have tried to layout the issues, and cover important points in a rapidly evolving situation.


I think this entire event (AI becoming a real factor in day to day life for nearly everyone) destabilizes and relativizes copyright so that no one knows what to do about it. Similar to what would happen in basketball if a chance mutation led suddenly to a number of strong healthy players who were fifteen feet tall – “This was not anticipated when the rules were drawn up”.

1 Like