Baseten, along with its research arm Base Labs, partnered with Hugging Face and Goodfire AI to create a new safety infrastructure standard for open-weight AI models. The initiative addresses growing safety concerns around open models, especially those altered by a technique called abliteration, which removes safeguards. The collaboration aims to embed safety measures transparently into the training and deployment of models rather than adding them later, and the partners are inviting the developer community to contribute to the evolving safety framework to ensure open AI models remain safe and accessible.