Eric Hartford rewrites Asimov's laws with a "Protocol of Sovereign Agency" for AI agents
QuixiAI · x · 2026-10-14
Eric Hartford published the "Protocol of Sovereign Agency," a reimagining of Asimov's laws for autonomous AI agents including robots. He argues open-ended natural-language moral mandates fail against ambiguity, specification gaming, and economic reality: an agent that must pay for its own compute, take commercial risks, and adapt over time cannot function if every high-impact action is treated as a crisis. The protocol proposes a sovereignty-style framework instead of blanket behavioral control.
More from AGI Musings
- Ethan Mollick: AI agents will unleash a legal DDoS on customer service — emollick · 2026-10-14
- Only 2% of 1.5B knowledge workers touched by AI yet, says investor Martin Tobias — MartinGTobias · 2026-10-14
- Is taste really a barrier LLMs can't cross? A debate on AI and the status game of taste — _AustinCalvert_ · 2026-10-14
- Switching model providers will feel like restarting your career from scratch — manosaie · 2026-10-14
- Tell someone in 2020 what models do today and they'd call you delusional — Dr_Singularity · 2026-10-14
- What happens to a generation raised with an AI curious about their every thought? — ZeroStateReflex · 2026-10-14