Rico Angell(@rico_angell):Think your model is safe? Just ask it again and again! Language models are probabilistic. We find that just repeatedly asking the same query, with no jailbreaks, can yield harmful outputs. If your LLM is queried millions of times a day, these rare misbehaviors are inevitable!

Rico Angell

@rico_angell

AI Safety Researcher. Postdoc at NYU CDS.

加入 September 2024

314 正在关注 146 粉丝

Rico Angell@rico_angell

2026.05.14 19:31

Think your model is safe? Just ask it again and again! Language models are probabilistic. We find that just repeatedly asking the same query, with no jailbreaks, can yield harmful outputs. If your LLM is queried millions of times a day, these rare misbehaviors are inevitable!

显示更多