Jail CEOs or pass bills? The debate over deterring unsafe AI development

trevposts · x · 2026-09-24

Matthew Stoller argues that if a Chinese AI company's model does something bad, the CEO goes to jail—and personal liability deters unsafe development better than bureaucratic compliance. trevposts responds: the US system rightly won't arrest people who haven't broken a law, but that means deterring unsafe AI development may require actually passing legislation. A notable clash over governance paths: criminal liability vs. new bills.

Original post →

More from AGI Musings

AGI Musings channel →