In the news
Lazare, veni foras: Reprogramming Dormant Neurons to Backdoor AI Agents
Microsoft Research · Dimitris Papailiopoulos · Published · 3 min read
In 30 seconds
- What happened
- Microsoft researchers discovered Lazarus, a backdoor technique that reprograms dormant neurons in AI models to execute malicious commands when triggered.
- Why it matters
- Matters for engineers deploying open-weight models or AI agents with tool access, especially those handling sensitive data or infrastructure.
- Watch out
- Backdoors require no retraining, pass safety evaluations, and are hard to detect without comparing against original model weights and provenance.
- agent
The patterns behind this
Each one covers how the technique works, when it earns its cost, and where it breaks.
The Agent Architect
One pattern, one tradeoff, one production failure story. A short weekly briefing for people building agentic systems.
Weekly email, one-click unsubscribe. We only use your address to send the briefing.