Christian Superintelligence is Faithful Order’s research program on the most capable AI systems: instruments that may one day match or exceed human performance across most intellectual work, and that must remain truthful, bounded, and subordinate to accountable human stewards under God.
Genesis places humanity in creation with a dual charge: to cultivate its latent possibilities and to keep and guard what God has made (Genesis 2:15). That charge holds even more strongly as computational tools grow more general and more consequential. When laboratories race to build systems intended to outperform humans across scientific, economic, and strategic work, two responses fail Christian duty alike.
The first failure is uncritical ambition: pursuing raw capability as a self-justifying end, speaking of machines in quasi-religious terms, or consolidating unaccountable power that answers neither to neighbor nor to God. The second failure is fearful withdrawal: burying technical gifts in the ground and leaving the moral ordering, evaluation, and governance of transformative tools entirely to purely commercial or materialist assumptions.
Christian Superintelligence exists to practice the third way set out in the Faithful Order Charter. Superintelligence, if it is ever achieved, will still be a creaturely artifact crafted by human hands. It will never possess a soul, moral rights, or spiritual authority. Precisely where capability is greatest, the obligations of truthfulness, ontological humility, human subordination, and preventive safety apply with the greatest force.
Our planned technical agenda addresses alignment in a single model, behavior across large interacting populations of agents, and stability across human-gated cycles of improvement.
We will post-train open-weight models to reason from within historic Christian faith while remaining rigorously truthful, fair to other traditions, and free of unsolicited preaching on ordinary tasks, verified by public evaluations and citation checks.
We will build isolated simulated communities of hundreds to thousands of interacting agents to test whether honesty, rule compliance, and distributed accountability hold at scale or erode into collusion, deception, and concentrated power.
We will study systems that propose improvements to their own training, evaluations, and tooling under one strict rule: every change passes automated checks and human doctrinal review before taking effect, with immediate halt authority at every step.
Christian Superintelligence operates under Faithful Order, governed by its Statement of Faith and subject to its Doctrinal & Stewardship Council. No research track here sits outside institutional oversight.
Every significant dataset, model training run, multi-agent simulation environment, and self-improvement experiment will undergo Pre-Project Stewardship Review before work begins and again before any external release. Simulations will run inside isolated environments, evaluations will be published so outside researchers can test our claims, and the Council holds binding authority to modify or halt any project regardless of competitive or financial cost.
Read more on our governance and review structure.
Faithful Order is newly founded, and we are assembling the evaluation suites, simulation protocols, and review structures described on this site. We welcome correspondence from machine-learning researchers, distributed-systems and security engineers, mathematicians, pastors, and theologians who share these convictions.
