Deep Ignorance: Filtering Pretraining Data Builds Tamper-Resistant Safeguards into Open-Weight LLMs

TL;DR

This paper demonstrates that filtering training data on dual-use topics during pretraining significantly enhances the resistance of open-weight LLMs to adversarial fine-tuning attacks, without impairing unrelated capabilities.

Contribution

It introduces a scalable multi-stage data filtering pipeline that effectively reduces biothreat knowledge in LLMs, establishing pretraining data curation as a key safety measure.

Findings

Filtered models resist 10,000-step adversarial fine-tuning

Filtering reduces biothreat knowledge without harming unrelated skills

Models can still access dangerous info via context, indicating a need for layered defenses

Abstract

Open-weight AI systems offer unique benefits, including enhanced transparency, open research, and decentralized access. However, they are vulnerable to tampering attacks which can efficiently elicit harmful behaviors by modifying weights or activations. Currently, there is not yet a robust science of open-weight model risk management. Existing safety fine-tuning methods and other post-training techniques have struggled to make LLMs resistant to more than a few dozen steps of adversarial fine-tuning. In this paper, we investigate whether filtering text about dual-use topics from training data can prevent unwanted capabilities and serve as a more tamper-resistant safeguard. We introduce a multi-stage pipeline for scalable data filtering and show that it offers a tractable and effective method for minimizing biothreat proxy knowledge in LLMs. We pretrain multiple 6.9B-parameter models from scratch and find that they exhibit substantial resistance to adversarial fine-tuning attacks on up to 10,000 steps and 300M tokens of biothreat-related text -- outperforming existing post-training baselines by over an order of magnitude -- with no observed degradation to unrelated capabilities. However, while filtered models lack internalized dangerous knowledge, we find that they can still leverage such information when it is provided in context (e.g., via search tool augmentation), demonstrating a need for a defense-in-depth approach. Overall, these findings help to establish pretraining data curation as a promising layer of defense for open-weight AI systems.