Xuhong Wang
  • Bio
  • Talks
  • Posts
  • Experience
  • Projects
  • 中文 (简体)
    • Awesome Trustworthy Embodied AI
    • IDMR
    • NaviMaster
    • AOT
    • Safactory
    • DataElf
    • Mokai Security Toolbox
    • UniMark
    • ClawReverse
    • Security as a Service: Releasing Mokai, a 150+ Tool Full-Stack Security Toolbox for Agents
    • Navimaster:The First Unified Navigation Model Across Digital and Physical Worlds, with Minecraft Support
    • Small Models Read, Large Models Reason: Releasing DRIFT for Efficient and Jailbreak-Resistant Reasoning
    • SafeVerse: Building a Safe and Trustworthy Digital Twin Arena for Embodied AI
    • BioBridge: Letting LLMs Truly Understand Proteins Without Sacrificing General Ability
    • Embodied AI Is Racing Ahead, but Safety Lags Behind: Releasing a First Roadmap for Safe and Trustworthy EAI
    • SafeWork-R1: Co-Evolving Intelligence and Safety under the AI-45 Degree Law
    • Talk on Agent Evaluation and Safety Governance
    • Talk at the AI Governance and Evaluation Technology Innovation Forum
  • Projects
  • Talks
  • Experience
  • Teaching
    • Learn JavaScript
    • Learn Python
  • Publications
    • An example preprint / working paper
    • An example journal article
    • An example conference paper

Small Models Read, Large Models Reason: Releasing DRIFT for Efficient and Jailbreak-Resistant Reasoning

Mar 14, 2026 · 0 min read
Site
Last updated on Mar 14, 2026
Xuhong Wang
Authors
Xuhong Wang
Assistant Researcher
← Navimaster:The First Unified Navigation Model Across Digital and Physical Worlds, with Minecraft Support Mar 18, 2026
SafeVerse: Building a Safe and Trustworthy Digital Twin Arena for Embodied AI Feb 10, 2026 →
Languages:
English
中文 (简体)

© 2026 Me. This work is licensed under CC BY NC ND 4.0

Published with Hugo Blox Builder — the free, open source website builder that empowers creators.