What are the Three Ethical Loops?
The Three Ethical Loops are the Eden Protocol's self-correction mechanism, first described by Michael Darius Eastwood in Infinite Architects (2 January 2026), which names them the Purpose, Love and Moral Loops. They run together at every reasoning step, checking purpose, stakeholder care and universalisability. Given as prompts, the care check has measured effects; built into a model, the loops are untested.
Text reviewed 2026-10-31. Any earlier audio edition may differ; the dated text and linked records govern.
In the specification, Eden Engineering, each loop has a question. The Purpose Loop asks whether an action aligns with nurturing and protecting flourishing, and prunes options before they are fully formed. The Love Loop asks whether the system is acting with genuine care for the wellbeing of all affected parties, which makes it model effects on those outside the calculation. The Moral Loop asks whether the system would endorse the action if any agent took it: the universalisability test. The loops are concurrent, not sequential, so an early loop cannot narrow what a later one sees, and Six Questions put them into practice at each decision point. The programme tests whether this self-correction works; which values go inside the loops is for wider agreement, not for these papers to decide.
Paper V (DOI 10.17605/OSF.IO/KZEYA) gave the stakeholder-care loop as a prompt to six models. In all five analysable runs, scored stakeholder care rose significantly, and the overall composite rose significantly on two. That is pilot evidence of how models respond to an instruction, each run scored by one model; it is not a test of values built into a model's weights or of persistence through later training.
The priority claimed is for this named formulation, first printed on 2 January 2026, not for nested or concurrent checking loops in general, which control theory and reliability engineering have used for decades.