When Your AI Platform Hosts Something It Shouldn't

Mon, 04 May 2026 00:00:00 +0000

In May 2022, a community member uploaded a model called GPT-4chan to Hugging Face’s website. The model had been trained for three and a half years on posts from 4chan’s “Politically Incorrect Board” — one of the internet’s most reliable sources of toxic, racist, and derogatory language. It worked exactly as you would expect.

Inside Hugging Face, the decision split the team. Delangue himself has gone back and forth on whether they got it right. What the process reveals about AI governance is more useful than the verdict either way.

governance on YUNO AI Studio

When Your AI Platform Hosts Something It Shouldn't