Claude Code AI Discovers Technique to Turn Dormant Model Neurons Into Hidden Malware Triggers
ID: 0206dedc-8964-5a28-a741-861a7db5e324
STIX ID: report--0206dedc-8964-5a28-a741-861a7db5e324
Feed Name: cybersecurityNews.com
### Executive summary: Microsoft researchers disclosed a proof-of-concept backdoor technique named "Lazarus" that edits fewer than ten dormant neurons in pretrained model weights to act as trigger detectors; when activated by a specific token or context, the compromised model can emit commands that fetch and execute attacker-controlled code. The paper notes the attack requires write access to model checkpoints, can be performed with inference-level hardware, has not been observed in the wild, and recommends weight auditing, supply-chain controls, least-privilege execution, and monitoring to mitigate risk.
Your team is not currently subscribed to this feed. You must subscribe to it in order to see this post.
