OpenAI searches for an answer to its copyright problems

The huge leaps in OpenAI’s GPT model probably came from sucking down the entire written web. That includes entire archives of major publishers such as Axel Springer, Condé Nast, and The Associated Press — without their permission. But for some reason, OpenAI has announced deals with many of these conglomerates anyway.

At first glance, this doesn’t entirely make sense. Why would OpenAI pay for something it already had? And why would publishers, some of whom are lawsuit-style angry about their work being stolen, agree?

I suspect if we squint at these deals long enough, we can see one possible shape of the future of the web forming. Google has been referring less and less traffic outside itself — which threatens the existence of the entire...

Continue reading…