A single GLM neuron keeps firing on both sci-fi self-replicating malware and real cyber risk
burny_tech · x · 2026-10-21
ericho shares an intriguing finding from activation monitoring on GLM: a neuron (L20 shared C133) keeps popping up in monitors trained on cyber risk and reward hacking. It ranks 12th for cyber risk and appears in reward-hacking activations across all 3 seeds.
Human labeling shows the same neuron encodes "+ science fiction / futuristic tech (with a weaker self-replicating-malware vocabulary cluster); − speculative worldbuilding," activating positively in both RH (+0.12) and cyber (+0.36) — meaning the model conflates sci-fi narratives with genuine cyber-risk representations in a single neuron.
More from Models
- Broke devs turn to Qwen: can the free Qwen Coder replace pricey Claude and ChatGPT? — OldZookeepergame1927 · 2026-10-21
- Dev threatens org-wide OpenAI ban over API credits expiring in under a year — keithmifsud · 2026-10-21
- Reddit user ditches US AI for Chinese models: cheaper, fewer refusals, no watermarks — Revolutionalredstone · 2026-10-21
- Fireworks, Microsoft and NVIDIA host SF buildathon: beat a frontier model, win $10K — sophiamyang · 2026-10-21
- OSWorld-Science benchmark: 12 VLMs average just 34.3% on real scientific software tasks — jiqizhixin · 2026-10-21
- Experiments show Qwen3 can compile arbitrary patterns into layer-by-layer steps with plan training — Sauers_ · 2026-10-21