Enhancing LLM Watermark Resilience Against Both Scrubbing and Spoofing Attacks

Open in new window