Steering Away from Memorization: Reachability-Constrained Reinforcement Learning for Text-to-Image Diffusion

Text-to-image diffusion models are susceptible to memorization, revealing a fundamental failure to generalize beyond the training set. Current mitigation approaches typically sacrifice image quality or prompt alignment to reduce memorization. To address this, we propose Reachability-Aware Diffusion Steering (RADS), an offline-trained, inference-time framework that mitigates memorization while maintaining generation fidelity. RADS models the diffusion denoising process as a dynamical system and uses reachability analysis to learn a safety value function that characterizes intermediate latent states along trajectories leading to memorized samples. This motivates a constrained reinforcement learning (RL) formulation, where a policy learns to steer the trajectory away from memorization via minimal perturbations in the caption embedding space. Empirical evaluations show that RADS achieves a stronger empirical Pareto frontier between generation diversity (SSCD), quality (FID), and alignment (CLIP) compared to state-of-the-art inference-time baselines. Crucially, RADS provides robust mitigation without modifying the diffusion backbone, offering a plug-and-play solution for diverse image generation.