Hugging Face is making an unusual move for a machine learning company: it’s openly documenting how it tries to put ethics into practice. The company just published the first edition of its Ethics and Society newsletter, which will appear quarterly at the equinox and solstice. Unlike many corporate ethics efforts that live in a siloed department, this one is assembled by the ‘Ethics and Society regulars’ — an open group of employees from across the company who work together as equals. The stated goal is to avoid the trap of having a dedicated team that lets everyone else off the hook. As the post puts it, making value-informed decisions requires ‘a shared responsibility and commitment from all parties involved.’
What’s striking is what the newsletter doesn’t do. Hugging Face offers no global list of values or principles, no top-down ethical framework. The team argues that ethics is too context-specific for that approach. Instead, they’re sharing project-specific thinking — documentation requirements, content guidelines, a Code of Conduct, and tools for community feedback on models, datasets, and Spaces. Anyone with Hub access can now provide direct feedback on hosted artifacts.
The timing matters here. The post notes that applied, tested practices for AI ethics have only emerged in the last 10 years, driven partly by ML models outgrowing the benchmarks used to measure them. As these systems moved into real-world applications affecting everyday life, the gap between theoretical ethics and practical implementation became impossible to ignore. Hugging Face’s open-source roots — collaboration, responsibility, transparency — put it in a different position than most companies now scrambling to retrofit ethics onto existing products.
Whether this approach scales remains an open question. A newsletter and a Code of Conduct don’t guarantee ethical outcomes, and the company is candid that this is an ongoing experiment rather than a solved problem. The real test will be whether these values survive contact with commercial pressure as Hugging Face continues to grow. But the decision to document the process in public, at a regular cadence, is itself a form of accountability that most AI companies have avoided.
