Self-Refining Language Model Anonymizers via Adversarial Distillation Paper • 2506.01420 • Published Oct 23, 2025