NUWA 
Deep Ignorance: Filtering Pretraining Data Builds Tamper-Resistant Safeguards into Open-Weight LLMs
QuestionCan withholding domain knowledge at pretraining provide a more durable safeguard than post-training alone?
Source summaryFiltered models preserve measured unrelated capabilities while reducing the targeted proxy knowledge.
NUWA research note · independent review pending · Contains: question / method / findings / interpretation / limits