Reprogramming dormant neurons to backdoor AI agents: Microsoft team's paper goes viral
DimitrisPapail · x · 2026-11-15
The paper "Lazare, veni foras: Reprogramming Dormant Neurons to Backdoor AI Agents" — by Yinzhi Cao, Giorgio Severi, Blake Bullwinkel, Eric Horvitz, Dimitris Papailiou and colleagues — has been certified as an "arXiv Banger," a community-recognized high-impact paper. It shows how dormant neurons in LLMs can be reprogrammed to plant backdoors in AI agents, a notable new study in agent security.
Related event: Reprogramming dormant neurons can backdoor AI agents without training(2 posts)→
More from Safety
- Editing Fewer Than 10 Dormant Neurons Can Backdoor AI Agents Without Any Training — DimitrisPapail · 2026-11-15
- Nvidia's security chief quietly left in August amid Huang fallout — experts say corporate security can't catch this kind of leak — TechCrunch AI · 2026-11-15
- LeCun vs Lambert: AI sovereignty favors open models, but that alone isn't enough — natolambert · 2026-11-15
- AI safety campaign calls for global ban on superintelligence and recursive self-improvement — marcvanderchijs · 2026-11-15
- Skillporter: a read-only-by-default MCP server to block skill injection — ImMandatory · 2026-11-15
- Zero Breach Toolkit v3: open-source home network defense built with Claude — PissraelExposed · 2026-11-15