Bottleneck for lab compute on critical infrastructure is willingness, not budget
herbiebradley · x · 2026-09-15
Responding to proposals that frontier labs allocate compute for public safety, herbiebradley argues the bottleneck for critical infrastructure work is willingness and logistics rather than budget: few plausible tasks are unconstrained by demand or could consume significant compute. Even securing a large fraction of open source software would likely be a small share of current compute. He finds the proposed 'allocate x% of compute to inference' strategy interesting overall.
More from Safety
- Beyond METR: 20 AI safety evaluators working with AI labs, from Apollo to Redwood — kevin_klyman · 2026-09-15
- Emad Mostaque warns METR and eval orgs will face nation-state level cyberattacks — QuixiAI · 2026-09-15
- Scholars blast arXiv for losing its way as a preprint server over AI crackdowns — RexDouglass · 2026-09-15
- FT editorial board endorses a pause on cutting-edge AI — peterwildeford · 2026-09-15
- Team kills AI security startup, open-sources local red-team engine OpenHunterAI — udmrzn · 2026-09-15
- Generative AI is the third information revolution after print and social media, says Prof. Nakov — preslav_nakov · 2026-09-15