Show HN: Distilling DeepSeek into GPT-OSS doesn't transfer censorship. Try it
Researchers investigated whether distilling knowledge from censored Chinese frontier AI models into American models transfers political censorship behaviors. The study found that while the student model successfully inherited advanced financial reasoning capabilities, it did not adopt the teacher's political biases or refusal patterns regarding sensitive topics.
Why it matters
This research suggests that organizations can leverage the high performance of foreign frontier models for specialized tasks without inheriting their restrictive censorship, potentially lowering costs and improving efficiency in AI development.
About Us Research FINANCE REQUEST ACCESS July 29, 2026 What a Distilled Model Inherits From Its Teacher Johnny Yu, Siddarth Mamidanna, Cyril Gorlla Introduction [ gpt-oss-20b-finance weights on Hugging Face ] [Try the playground ] [ LineageEval ] [Explore the data on GitHub ]
Get smarter about the news
Sign up free for a feed built around what you actually care about, Dive Deeper research on any story, and the full text of every article.
Create free accountAlready have an account? Sign in