给AI装个安全开关：想禁什么，拔掉那根线就行
现在的AI画图模型，要么喂脏数据(什么都能画，但也会画不该画的)，要么只喂干净数据(安全了，但画得也差了)。这篇论文换了个思路：不删数据，而是在模型内部把「安全知识」和「不安全知识」物理隔开，像电路板上的两根独立线路。训练时一起学，生成时把「不安全」那根线直接拔掉。在ImageNet和更大的CC12M数据集上，效果比只喂干净数据或事后删除概念都好。它不是你明天能用上的，但「把知识分区、按需断电」这个思路，可能比现在所有「事后补救」的安全方案都更接近问题的根。
📄 原文摘要(英文)
Training on broad, unfiltered data enables high-quality image generation, but can also teach models concepts that should not be generated; restricting the training data ensures safety, but reduces quality because examples containing unsafe concepts also provide useful knowledge that is not itself unsafe. Existing approaches largely address this trade-off post hoc through concept unlearning, which can lead to catastrophic forgetting as the number of concepts to remove grows. We instead ask whether unsafe concept knowledge can be compartmentalized during pretraining so that access to it can be disabled at inference. We introduce Pristine Models, a novel text-to-image generation architecture with an attention mechanism that encourages the separation of safe and unsafe knowledge into independently parameterized branches. The branch-specific parameters are learned during training, while the unsafe branch can be detached at inference. On ImageNet, Pristine Models enable safer generation while retaining much of the utility of full-data pretraining, achieving higher image quality and better text alignment than training only on safe data or applying post-hoc concept removal. We further scale our approach to CC12M, demonstrating that it remains effective in a substantially larger and more diverse open-world pretraining setting.