ONTO · Archive
Every result is a document.
Numbered and dated, like a preprint server. Published texts are never rewritten; corrections come as new versions. Closed versions keep their date and fingerprint only.
29documents
Discipline in the weights
Stage Beta · 5ONTO:2610.00042026-10-07How ONTO teaches and checks a modelOpen · v2ONTO:2610.00052026-10-10Everyday questions: the student against its base✓ OpenONTO:2610.00062026-10-05Stage Beta acceptance: the exam machine and its first verdict✓ OpenONTO:2610.00072026-10-10How ONTO teaches and checks a model: methodology v5✓ OpenONTO:2610.00082026-10-10Stage Beta, cycle by cycle✓ Open
The institute
Prompt layer, February–April 2026
These documents measured a canon delivered through the system prompt of production models and scored by the engine of that period. They are not results of the current method. Published texts are never rewritten: where content had to change, a version 2 lists every change, and version 1 keeps its date and fingerprint. What we no longer claim · Why we left the prompt layer.
Documents of the period · 9ONTO:2601.00022026-01-27ONTO Epistemic Risk Standard, version 1.0ONTO-ERS-1.0✓ OpenONTO:2603.00072026-03ONTO Standard: The Operating System for AI Quality (whitepaper)WP-2026-002Open · v2ONTO:2603.00082026-04Methodology of the prompt layer (from the documentation)Open · v2ONTO:2603.00092026-04ONTO Epistemic Risk Standard v10.0 with regulatory alignment✓ OpenONTO:2603.00102026-04The problem and the roadmap, spring 2026Open · v2ONTO:2603.00112026-04Governance ONTO-GOV-001✓ OpenONTO:2604.00012026-04-20Formal foundation of IGR (from the pitch, v5.13)✓ OpenONTO:2604.00022026-04-25Regulatory frameworks and 14 countries, April 2026✓ OpenONTO:2604.00032026-04Commercial pages of April 2026✕ Closed
100-question experiment · 4ONTO:2602.00012026-02-14GOLD Experiment Questions v1.0DS-2026-001✓ OpenONTO:2602.00022026-02-14ONTO-GOLD BASELINE ANALYSIS REPORTCS-2026-001Open · v2ONTO:2602.00032026-02-14ONTO-GOLD VALIDATION REPORT: BEFORE → AFTERER-2026-001Open · v2ONTO:2602.00042026-02-14ONTO-GOLD: BEFORE → AFTER — FULL TEXT COMPARISONER-2026-002Open · v2
Models in the field · 3ONTO:2602.00052026-02-19ONTO Field Observation FO-2026-003FO-2026-003✓ OpenONTO:2603.00032026-03-07FO-2026-004: Spontaneous Epistemic Drift in Claude Haiku 4.5 — Longitudinal ObservationFO-2026-004Open · v2ONTO:2603.00042026-03-08CS-2026-002 — Baseline Epistemic AssessmentCS-2026-002✓ Open
Proxy demonstrations · 5ONTO:2603.00012026-03-03F-001: Grok vs Grok+GOLD — "AGI within 10 years?"F-001Open · v2ONTO:2603.00122026-03-03I Tested ONTO Standard. Here's What I Found.IR-2026-001✕ ClosedONTO:2603.00022026-03-04F-002: Are Statins Effective for Primary Prevention?F-002Open · v2ONTO:2603.00052026-03-13CS-2026-003: Naive Question Epistemic Analysis - Grok 4CS-2026-003Open · v2ONTO:2603.00062026-03-13CS-2026-004: Naive Question Epistemic Analysis - GPT (OpenAI)CS-2026-004Open · v2