tszzl warns open models nearing superintelligence: offense dominates defense, US and China will intervene
JasonBotterill · x · 2026-09-13
AI researcher tszzl says he loves open models (running a Kimi K3 finetune on tinker) but warns that free distribution of open-source models approaching superintelligence could go badly: "the offense dominates the defense," and both China and the US would then try to control it. Replies raise whether labs would cap inference speed (e.g., no Cerebras-level tokens/s on frontier models) as a capability control.
More from AGI Musings
- Domingos: we should celebrate AI and math progress, instead we're in collective neurosis — pmddomingos · 2026-09-13
- Lab leaders genuinely believe in AGI risk while racing, author argues — menhguin · 2026-09-13
- AI labs' slowdown pledges split the internet: safety sincerity vs regulatory capture debate — menhguin · 2026-09-13
- Pedro Domingos: Apocalypticism baked into Western psyche is a civilizational weakness — pmddomingos · 2026-09-13
- Domingos cites cybersecurity experts: AI hasn't done anything we couldn't already do — pmddomingos · 2026-09-13
- Domingos: I'm more worried Claude is aligned with Anthropic than not aligned — pmddomingos · 2026-09-13