Want to create an interactive transcript for this episode?
Podcast: Arxiv Papers
Episode: [QA] Deep Ignorance: Filtering Pretraining Data Builds Tamper-Resistant Safeguards into Open-Weight LLMs
Description: This paper explores filtering dual-use topics from training data to enhance the tamper-resistance of open-weight AI systems, demonstrating significant improvements in adversarial fine-tuning resistance without degrading unrelated capabilities.https://arxiv.org/abs//2508.06601YouTube: https://www.youtube.com/@ArxivPapersTikTok: https://www.tiktok.com/@arxiv_papersApple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id1692476016Spotify: https://podcasters.spotify.com/pod/show/arxiv-papers