AI Safety on a New Scale: Can Transparency Succeed Where Regulation Fails? Base Labs, the research arm of Baseten, has partnered with Hugging Face and Goodfire to develop an open weight AI safety partnership.
The partnership aims to create a standard for transparency and monitoring in AI development by developing methods for training and publishing open weight models that are inherently transparent.
Over 6,000 AI models hosted on Hugging Face have been "ablitreated," or had safeguards removed, rendering them potentially perilous.