Safety researcher: no GLM 5.3 cyberattack wave doesn't refute risk concerns
dhadfieldmenell · x · 2026-10-09
AI safety researcher dhadfieldmenell pushes back on claims that GLM 5.3's release without a wave of cyberattacks disproves risk worries. He concedes the point has some merit but argues these capabilities will diffuse into more efficient models and deployments, making skeptics look silly later.
His core argument: the field should focus less on whether openly released weights expose a capability, and more on how cheap and accessible it is to actually use that capability.
Related event: Researcher Pushes Back on Claims That Open-Model Risks Are Overblown(2 posts)→
More from AGI Musings
- DeepMind essay: what 15M Gemini interactions reveal about AI-accelerated science — JMateosGarcia · 2026-10-09
- AI Swarmhood instead of Personhood: rethinking how we frame machine minds — theimposingshadow · 2026-10-09
- World models, not LLMs, are the bridge to physical intelligence — ingliguori · 2026-10-09
- Dietterich: treat OpenAI's math models like a deceased mathematician's notebook — tdietterich · 2026-10-09
- The Best Enterprise AI Opportunities Hide in Undigitized Workflows — alex_verem · 2026-10-09
- 45 people have left frontier AI labs citing safety and ethics concerns — birchlse · 2026-10-09