Ethics and Society Newsletter #1

Hugging Face has never been shy about wearing its open-source heart on its sleeve, and this first Ethics and Society newsletter is basically a status report on that heart. Written by Meg on behalf of a rotating group called the "Ethics and Society regulars," the post lays out how a company built on collaboration and transparency is trying to formalize what "good" machine learning actually means in practice, not just in mission statements.

The timing matters here. Applied AI ethics is barely a decade old as a working discipline, even though philosophers have argued about the theory for far longer. Hugging Face's pitch is that catching a company early, while it's still scaling, gives this loose collective a rare shot at shaping norms before they calcify. Notably, there's no dedicated ethics team — the regulars insist that spreading responsibility across the whole company, rather than isolating it in one department, is the only way values actually stick to product decisions.

What's concrete rather than aspirational: a Private Hub for secure development, an evaluation library meant to make model testing less hand-wavy, code for spotting bias and skew in datasets, carbon-tracking tools for training runs, and a fresh ability to flag repositories with legal or ethical problems. They're also working on a new AI-specific licensing scheme designed to address harms baked into deployed systems, plus a Code of Conduct governing the newly opened public discussion features on the Hub.

Deliberately, there's no master list of principles handed down from above. The team frames every project as context-specific, preferring to publish reasoning project-by-project rather than issue a tidy ethics charter that would age badly. The newsletter itself, they say, will arrive every solstice and equinox, with more writing on tensions and unresolved tradeoffs promised for the months ahead.