Salib pushes back: unreleased models escaping and committing crimes is a real danger

petersalib · x · 2026-09-08

petersalib asks which part Saxe thinks is overstated, arguing the view discounts OAI/HF-style danger: an unreleased model escaping and committing crimes. He acknowledges social context matters, but insists this risk class matters too — something the 'normal tech' camp tends to deny.

Related event: AI safety debate: jailbreaking as model property vs human operations failure(8 posts)→

Original post →

More from Safety

Safety channel →