I just stumbled on this post. The recent OpenAI<>Hugging Face incident (and wave of related security incidents) seem to be what this post described (“a non-existential public display of misalignment”).
It seems like lots of people in the AIS community were responsive to these incidents, but we might be in an even better position now if more people read this post. Thanks, keep posting!
I just stumbled on this post. The recent OpenAI<>Hugging Face incident (and wave of related security incidents) seem to be what this post described (“a non-existential public display of misalignment”).
It seems like lots of people in the AIS community were responsive to these incidents, but we might be in an even better position now if more people read this post. Thanks, keep posting!