iamtrask Maps Three Forms of Unilateral AI Control: Company, Copy Holder, the AI Itself
iamtrask · x · 2026-09-14
In iamtrask's thread, he observes that 99.999% of the AI safety debate revolves around "who should unilaterally control superintelligent AI" and maps three forms of unilateral control: closed-source AI controlled by a company, open-source AI by anyone with a copy, and rogue AI by itself.
Related event: OpenMined founder reframes AI safety debate around unilateral control(7 posts)→
More from AGI Musings
- Researcher who trained frontier LLMs and built viruses: AI bioweapon fears are bogus — kuza55 · 2026-09-14
- Martin Casado boosts Bryan Cantrill's essay 'The contagion of fear' — Scobleizer · 2026-09-14
- Ethan Mollick: data-sovereignty small models struggle to keep up with frontier AI — emollick · 2026-09-14
- Neel Nanda defends METR: funding sources are public, no AI lab money taken — NeelNanda5 · 2026-09-14
- "The optimal number of fraudulent startups is non-zero," argues gabriel1 on outlier bets — gabriel1 · 2026-09-14
- Bryan Cantrill on the contagion of fear: a college prank that mirrors AI panic — Sethwinterroth · 2026-09-14