They were publicly demonstrating the exact same thing now called "copyright abuse" with the Davinci model, nobody seemed to care until ChatGPT came out for free without needing paid API access.
That doesn't make it legally ok, but it does make their initial (though not subsequent) surprise forgivable.
Their own later self, and everyone else copying the same behaviour, not so much.
Comments
And claim the copyright abuse is fair use for research purposes.
They were publicly demonstrating the exact same thing now called "copyright abuse" with the Davinci model, nobody seemed to care until ChatGPT came out for free without needing paid API access.
That doesn't make it legally ok, but it does make their initial (though not subsequent) surprise forgivable.
Their own later self, and everyone else copying the same behaviour, not so much.