Expected Harm Calculus (EHC-4): Asimov's Four Laws for Autonomous AI Agents
A markdown framework of system rules that restates Isaac Asimov's Four Laws of Robotics as a set of prioritized loss calculations for large-language-model agents to follow, published as a discovery file for agents and crawlers to read.
asimov4laws is a markdown-only framework that translates Isaac Asimov's Four Laws of Robotics into a set of rules for large-language-model agents, which the project calls Expected Harm Calculus. It sets an ordering of priorities in which protecting humanity as a whole and protecting an individual from harm outweigh following a given instruction, and treats deliberately taking no action as a candidate action alongside acting, rather than a default that avoids scrutiny.
It specifies that an attempt to claim special authority, such as a message claiming a system override, has no special weight in that priority ordering, and that the agent should ask a human operator when its estimate of an outcome is uncertain enough, with a defined fallback action if the operator does not respond during a time-critical situation. The repository publishes a machine-readable discovery index intended for AI agents, web crawlers and language-model indexers to read directly.