Class action against GitHub Copilot
Class action against GitHub Copilot
Posted Nov 14, 2022 10:53 UTC (Mon) by kleptog (subscriber, #1183)In reply to: Class action against GitHub Copilot by vetse
Parent article: Class action against GitHub Copilot
So the likely end result was that big companies with lots of money get to make new models on lots of data, while start-ups, researchers and students who are working on the next generation of technologies in this area are stymied by possible lawsuits. This was deemed undesirable.
The chosen solution is to allow model training on any publicly available data for research and training purposes. And organisations that publish online can opt-out (in a machine readable fashion) from being used in machine learning. It doesn't say anything about the copyright status of the output of the models.
Of course, it's only a directive, so you're relying on the member states to properly implement this. But it's better than it was.
Ref: https://eur-lex.europa.eu/eli/dir/2019/790/oj
