Microsoft Research shows dormant-neuron backdoor can hide in AI agents

1. 6:18 PM
   Microsoft Research@MSFTResearchOfficial

   Microsoft Research shows a dormant-neuron backdoor can hide in AI agents

   AIMicrosoft Research demonstrates that reprogramming just a few dormant neurons in an AI agent can create a backdoor. The agent behaves normally until it encounters a specific trigger. Microsoft says the mitigation is to verify where a model came from and limit its permissions, and it points to Microsoft Foundry safety systems for Azure Direct Models and Hugging Face models.

   
2. 3:52 PM
   Microsoft ResearchOfficialPick

   Microsoft Research discloses Lazarus, a dormant-neuron backdoor for AI agents

   AIMicrosoft researchers disclosed Lazarus, a backdoor technique that reprograms fewer than 10 dormant neurons in an LLM to make an agent run attacker commands when a hidden trigger appears. The researchers say the method requires no training, the model passes standard evals until the trigger, and the agent then runs the command with whatever access it has. They recommend verifying model provenance, auditing weights against originals, limiting agent permissions, and monitoring agent actions.

The trend compares only continuously observed participants, so its scope may be smaller than the current heat total. Hover or tap for hourly heat; use arrow keys with a keyboard.