Baseten's Base Labs, Hugging Face, Goodfire to build open-weight safety standard

Baseten is not an unknown player in the AI infrastructure space. The company, an inference provider, raised a $1.5 billion Series F in June, vaulting its valuation to $13 billion, and spun up the Base Labs research group earlier this year. Its partner Goodfire AI is similarly well-capitalized, having raised a $150 million Series B led by B Capital earlier this year to advance its model interpretability platform. That financial backing suggests these are not fledgling experiments but well-funded efforts to shape how open models are trained and deployed.

The safety question around open-weight models has become more urgent as abliteration spreads. Hugging Face, which hosts open source AI models, currently lists over 6,000 abliterated models, a concrete measure of how readily safeguards can be stripped away. Baseten is framing its work as a \"standard\" for open models that is transparent and built into training and deployment, rather than bolted on afterward. In a post on X, the company said it believes openness is an advantage for AI safety, providing more visibility into model behavior and greater means of turning safety research into actionable controls than closed-source approaches.

Goodfire, which specializes in opening AI's \"black box\" to explain how models make decisions, is the likeliest candidate for the \"built into\" part of that equation, given its interpretability focus. Baseten is also putting out an open call to the broader developer ecosystem to contribute to the framework, saying it is building an ecosystem of open models that are safe and accessible to all. Whether this becomes a widely adopted standard or remains a well-funded but niche effort will likely depend on how many developers answer that call.