The Hugging Face incident and impacts of misaligned models on third parties
openai.com
1 thread
This incident feels like a marketing stunt. Any company with the skill to build a frontier-level model, also has the skill to contain it until it is ready to share.
apparently they dont have both skill sets, but could certainly afford to
likely a managerial problem at the core
what would happen if they supplement the IT team using AI to show them how to build better sandboxes. Do you think it would be a conflict of interest for the AI?