The Trojan Knowledge: Bypassing Commercial LLM Guardrails via Harmless Prompt Weaving and Adaptive Tree Search
- Rongzhe Wei
- Peizhi Niu
- et al.
- 2026
- ICML 2026
Conference paper
Publications
38 results at ICML 2026
Towards Effective Waste Segmentation for Automated Waste Recycling in Cluttered Background
- Mamoona Javaid
- Mubashir Noman
- et al.
- 2026
- ICML 2026
Conference paperUnlocking Speech–Text Compositional Powers: Instruction-Following Speech Language Models without Instruction Tuning
- Congrui Du
- Yang Zhang
- et al.
- 2026
- ICML 2026
Conference paperPushing Forward Pareto Frontiers of Proactive Agents with Behavioral Agentic Optimization
- Yihang Yao
- Zhepeng Cen
- et al.
- 2026
- ICML 2026
Conference paperPRISM: Demystifying Retention and Interaction in Mid-Training
- Bharat Runwal
- Ashish Agrawal
- et al.
- 2026
- ICML 2026
Conference paperPosition: Agentic Systems Should be General
- Elron Bandel
- Asaf Yehudai
- et al.
- 2026
- ICML 2026
Conference paperCapability-Oriented Training Induced Alignment Risk
- Yujun Zhou
- Yue Huang
- et al.
- 2026
- ICML 2026
Conference paperOne Bug, Hundreds Behind: LLMs for Large-Scale Bug Discovery
- Qiushi Wu
- Yue Xiao
- et al.
- 2026
- ICML 2026
Conference paperVividCam: Learning Unconventional Camera Motions from Virtual Synthetic Videos
- Qiucheng Wu
- Handong Zhao
- et al.
- 2026
- ICML 2026
Conference paperPosition: Collusion Risks Among AI Reasoning Agents Justify Certification Requirements for Making Market Decisions
- Matthew Riemer
- Tommaso Tosato
- et al.
- 2026
- ICML 2026
Conference paper