{"id":"arxiv/2606.04775","url":"https://arxiv.org/abs/2606.04775","tags":["cs.LG","cs.AI","cs.CV","eess.SY","math.OC"],"score":0,"title":"Activation Steering of Video Generation Models via Reduced-Order Linear Optimal Control","domain":"cs.LG","source":"arxiv","authors":["Jihoon Hong","Alice Chan","Qiyue Dai","Julian Skifstad","Glen Chou"],"reviews":0,"subject":"cs.LG","summary":"Text-to-video (T2V) models trained on large-scale web data can generate undesired content, motivating interventions that reduce harmful outputs without sacrificing visual quality. Activation steering offers an attractive mechanistic alternative to finetuning and prompt filtering, but existing T2V steering methods remain limited, typically applying coarse, non-anticipative interventions that can lead to oversteering and content degradation. To close this gap, we propose Latent Activation Linear-Quadratic Regulator (LA-LQR), a reduced-order optimal control framework for minimally invasive T2V…","category":"research","comments":0,"disputed":false,"division":"cs","mentions":0,"discussed":0,"published":"2026-06-03","reactions":{}}