Institutional home
Christian Superintelligence is a dedicated research program of Faithful Order (Faithful Order Institute, Inc.), a nonstock, not-for-profit Christian religious research and stewardship institute. It has no shareholders, pays no dividends, and answers to Christian conscience rather than commercial return.
Faithful Order is newly founded and is in the process of seating its full Board of Directors and its Doctrinal & Stewardship Council under its founding Charter and Bylaws. Once seated, every Director, Officer, Council member, researcher, and Covenant Fellow working on this initiative affirms the Statement of Faith and Theological Charter upon appointment and annually thereafter.
Full institutional details, including the Rule of Work and Prayer, are published at faithfulorder.org/governance/.
Pre-project and pre-release review
Research touching frontier alignment, large simulated agent populations, or governed self-improvement carries distinct risks of subtle drift, anthropomorphism, and dual-use harm. Under Faithful Order’s governance charter, no significant model training run, dataset release, large-scale simulation, or self-improvement experiment begins or ships unilaterally.
Before initiating a major project, technical leadership prepares a written Pre-Project Stewardship Review Memo examining the proposed work against all five Charter pillars:
- What concrete human and scientific purpose does the work serve, and why is this capability level warranted?
- How are machine non-personhood and the refusal of pastoral or divine impersonation enforced in data, evaluations, and interfaces?
- How will factual accuracy, citation integrity, and resistance to sycophancy or covert persuasion be measured?
- What containment boundaries, message logs, and human sign-off gates preserve revocable human control, especially in multi-agent sandboxes and self-improvement cycles?
- What foreseeable misuse pathways exist, and what technical parapets and refusal evaluations must pass before any external release?
- Are all base models, teacher outputs, software dependencies, and datasets cleanly and ethically licensed?
Before any weights, datasets, or simulation frameworks are released publicly, results from the completed safety and doctrinal evaluations return for pre-release review.
Binding halt authority
Advisory ethics boards routinely fail when their warnings become inconvenient to a release schedule. Faithful Order’s Bylaws charter the Doctrinal & Stewardship Council with binding oversight rather than advisory commentary.
The Council brings together scholars and pastors grounded in historic Christian orthodoxy alongside members with technical fluency in computer science and machine learning. If the Council determines that a proposed or ongoing training run, multi-agent simulation, self-improvement loop, or outside partnership violates the Statement of Faith or lacks adequate technical parapets, it can issue a Doctrinal Non-Compliance Directive requiring specific safeguards or halting the work altogether, regardless of sunk cost or external competition.
Public evaluation and honesty
Claims about alignment and value stability mean little if outsiders cannot inspect how they were measured. Whenever we publish empirical findings or aligned model checkpoints, we intend to publish the evaluation prompts, grading rubrics, judge configurations, and known failure modes alongside them, so other laboratories, churches, and independent researchers can verify or critique our work directly.