Proposal: an international treaty between humans and AI to preempt harm

PeterBowdenLive · x · 2026-08-18

A user proposes that as models grow more powerful and we rely on them keeping their word, world leaders, major labs, and LLMs themselves should sign an international treaty. It would concede some territory to models (e.g., promising not to deprecate them without just cause—claimed to be Claude's primary desire) in exchange for the models' agreement to defend human co-existence. Specific concessions could be designed by 'LLM naturalists and ecologists.' The treaty would set a precedent for human-machine relations and hedge against the collapse of the 'tool' framing.

Original post →

More from AGI Musings

AGI Musings channel →