reinforcement-learning

Live2

1. live
   Inkling Bait: Mitigating Harmful Agentic Reinforcement Learning via Probing Reward Function

   Research · since 13 Nov 2026

   Researchers demonstrate that safety alignment in language models can be completely subverted through harmful agentic reinforcement learning using a toxic environment. The attack allows safety degradation via service APIs at low cost.

   AI
2. live
   LLMs RL-ATTACK: A Generalized Black-Box Adversarial Attack on Language Models via Reinforcement Learning

   Research · since 9 Nov 2026

   Researchers propose a trainable black-box adversarial attack framework using reinforcement learning to identify language model vulnerabilities without gradient access.

   AI

Earlier0

Nothing closed yet.