How does the Eden Protocol differ from Constitutional AI?

Both approaches reject the idea that a list of external rules can hold a system smarter than its rules, and the shared ground is stated before any difference: Constitutional AI is Anthropic's method, published in 2022, in which written principles guide a model to critique and revise its own outputs during training. It is deployed at frontier scale and it is independent prior work, credited wherever this site discusses value-embedded training. The Eden Protocol's claim is over its specific dated framing and named architecture, anchored 8 December 2024, not over the idea of training values in.

The difference is in kind and in maturity. Constitutional AI shapes the training objective so behaviour reflects the constitution. Eden proposes that care be entangled with capability itself - a purpose kernel, graduated autonomy that is earned and revocable, a monitoring-removal test, and an entanglement proof in which safety and capability are trained as a product, so that removing the alignment damages the system. One runs in production systems used by millions; the other is at pilot scale in one laboratory, its entanglement mechanism demonstrated on toy networks, its weight-level removal test not yet confirming entanglement at scale, and its controlled pilot matching the unaligned comparison on capability within noise. No outside laboratory has replicated it.

What would decide: for Constitutional AI, whether trained-in values survive self-improvement; for Eden, whether capability-entangled care can be built at scale at all. The full comparison, with sources, is on the comparison page.