LLMs Safety

Robust Safety Monitoring of Language Models via Activation Watermarking
Robust Safety Monitoring of Language Models via Activation Watermarking

A robust monitoring framework for LLM safety that uses activation watermarking against adaptive attackers.

Mar 24, 2026

Towards Understanding the Fragility of Multilingual LLMs against Fine-Tuning Attacks
Towards Understanding the Fragility of Multilingual LLMs against Fine-Tuning Attacks

Fine-tuning attacks in one language can break multilingual safety alignment, suggesting safety information is partly language-agnostic.

Oct 1, 2024