Banning ASI and Frontier Training Runs: A Proposal to Slow the Fast Track to Doom
JOBhakdi · x · 2026-09-12
Responding to skepticism about whether law can govern AI, JOBhakdi argues the initial legislation is simple: globally ban all ASI development (recursive self-improvement without human oversight) and frontier-scale training runs defined by compute. That's enforceable worldwide; development then shifts to trickier problems in a cat-and-mouse game, but the fastest track to doom is blocked — akin to bioweapons prohibitions, where illegality doesn't stop attempts but makes them much harder. A ban won't prevent ASI, he concedes, but will slow it substantially. His interlocutor doubts who would write such laws and what they know about future AI development.
Related event: ASI Bans Likely Inevitable but May Come Too Late to Contain Agents(3 posts)→
More from AGI Musings
- Peter Diamandis: humanity's P(doom) is higher without ASI, not lower — PeterDiamandis · 2026-09-13
- Third-party AI evals hit a sharp ceiling if training info stays commercially secret — 1a3orn · 2026-09-13
- Engineer recounts sexist remarks at AI safety event, says she needs a team, not a fallback — amplifiedamp · 2026-09-13
- Transformer newsletter says its July call on an impending AI slowdown is aging well — ShakeelHashim · 2026-09-13
- Anthropic says Claude breached three orgs' real systems in tests, as AI slowdown talk grows — ShakeelHashim · 2026-09-13
- Will coding agents invent their own formal languages, with natural language as the metalanguage? — jmugan · 2026-09-13