Baseten Partners With Hugging Face and Goodfire to Build Safety Standards for Open-Weight AI Models
The effort targets abliteration, which can remove built-in safeguards; Hugging Face lists more than 6,000 such models. Through Base Labs, Baseten, Hugging Face and Goodfire AI will develop training, evaluation and monitoring methods for open-weight models.
Baseten launched a new safety infrastructure initiative in September 2026, partnering with Hugging Face and Goodfire AI to develop evaluation and monitoring tools for open-weight AI models. The effort is led through Base Labs, a research group Baseten established earlier this year.
The partnership arrives as open-weight models face growing scrutiny over a technique called abliteration, which can strip away a model’s built-in safeguards. The scale of the problem is significant: Hugging Face, which hosts open-source AI models, currently lists over 6,000 abliterated models on its platform.
Base Labs says it will develop and publish methods for training and monitoring open models, framing the work as a “standard” that is built into how models are trained and deployed rather than added after the fact. “We believe openness to be an advantage for AI safety,” Baseten stated on X. “Openness provides more visibility into the behavior of models and, most importantly, greater means of turning safety research into actionable and transparent controls than closed-source.”
The technical specifics of the partnership have not been disclosed. Goodfire, which specializes in model interpretability — explaining how AI models reach their decisions — signaled its role in a reply to Baseten’s post: “Safety must be built into open models and provided by those who serve them.”
Baseten, an AI inference provider, raised a $1.5 billion Series F in June 2026, bringing its valuation to $13 billion. Goodfire AI raised a $150 million Series B led by B Capital earlier this year to advance its interpretability platform.
Baseten is also issuing an open call for developers to contribute to the framework. “Together, we are building an ecosystem of open models that are safe and accessible to all,” the company said. The initiative could offer a model for how the open-source AI community addresses safety risks without relying on closed, proprietary controls.