[newhaus1994]

In spite of all that we're studying wrt abuse potential, I (and my team) generally support open-sourcing tech, and I hope that we can contribute not to "oh this is dangerous, don't release" but rather to "oh this is dangerous, it's already released, what are we going to do now?"

[40four]

Great, keep up the good work! Are you able to discuss how studies like yours work? Is it along the lines of determining if people can distinguish between human written and AI generated text? Sounds like a difficult question to answer.

I suspect they will release the full model in time. It's already trending in that direction.