Nathan Lambert: the open-weights cyber weapon scare hasn't materialized—GLM-5.3 proves the point
natolambert · x · 2026-10-31
Nathan Lambert responds to critics of his open-models-and-cyber-risk take, arguing the AI community is asking the wrong question—not which model type causes harm, but how cybersecurity maintains equilibrium and who wields new tools.
Key points: recent mass vulnerability disclosures (PHP2Shell, GitLab) got dubbed "Black October," signaling AI-enabled cyber risk materializing; open-weight Claude was previewed in April as a potential cyber weapon that could destabilize society—by his measures that prediction failed; GLM-5.3, the model actually crossing the capability threshold, has been out over a month with little evidence of changed harm; WSJ reports cyber first responders are strained, which he reads as the system working through an exponential vulnerability curve that will take a long time to level off.
More from Models
- ThursdAI crew debates: with Opus 7 coming, who still cares about better benchmarks? — thursdai_pod · 2026-10-31
- 14 models tested on 960 analyst tasks: AI scores 97/100 as analyst, picks winners no better than random — HeyAmit_ · 2026-10-31
- Anthropic adds invisible text watermark to Claude outputs starting Nov 3 — petesena · 2026-10-31
- GPT-6 Luna vs GPT-6.1 Terra vs GPT-6.1 Sol vs GPT-6 Astra in Pixel Art — Angaisb_ · 2026-10-31
- Perplexity engineer invites public tasks to prove or fix pplx-decider's SOTA claims — antoine_chaffin · 2026-10-31
- Arena weekly: Gemini 4 Carbon tops Text Arena, GPT-6.1 Terra enters Agent Arena at #8 — arena · 2026-10-31