Machine Learning Security

Mitigating Watermark Forgery in Generative Models via Randomized Key Selection