Subversion Strategy Eval: Can language models statelessly strategize to subvert control protocols?by Alex Mallen, Charlie…
Exploring Memorization and Copyright Violation in Frontier LLMs: A Study of the New York Times…
The BrowserGym Ecosystem for Web Agent Researchby Thibault Le Sellier De Chezelles, Maxime Gasse, Alexandre…
Frontier Models are Capable of In-context Schemingby Alexander Meinke, Bronson Schoen, Jérémy Scheurer, Mikita Balesni,…
Best-of-N Jailbreakingby John Hughes, Sara Price, Aengus Lynch, Rylan Schaeffer, Fazl Barez, Sanmi Koyejo, Henry…
LMAct: A Benchmark for In-Context Imitation Learning with Long Multimodal Demonstrationsby Anian Ruoss, Fabio Pardo,…
LLMStinger: Jailbreaking LLMs using RL fine-tuned LLMsby Piyush Jha, Arnav Arora, Vijay GaneshFirst submitted to…
NeKo: Toward Post Recognition Generative Correction Large Language Models with Task-Oriented Expertsby Yen-Ting Lin, Chao-Han…
TableGPT2: A Large Multimodal Model with Tabular Data Integrationby Aofeng Su, Aowen Wang, Chao Ye,…
Sabotage Evaluations for Frontier Modelsby Joe Benton, Misha Wagner, Eric Christiansen, Cem Anil, Ethan Perez,…