Saved in:
| Main Author: | |
|---|---|
| Format: | Recurso digital |
| Language: | English |
| Published: |
Zenodo
2026
|
| Subjects: | |
| Online Access: | https://doi.org/10.5281/zenodo.18969708 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866901620980187136 |
|---|---|
| author | Mariquit, Erny-Jay |
| author_facet | Mariquit, Erny-Jay |
| contents | <p>We present TSCWH (The System for Covenant-Weighted Heuristics), a patent-pending AI safety evaluation architecture achieving zero-LLM-call deliberative consensus through a proprietary shared-memory coordination design.</p> <p>KEY RESULTS:<br>• 10 agents deliberate across 8 ethical dimensions (Charity, Grace, Stewardship, Truth, Dignity, Courage, Community, Creation Dignity) in <50 ms<br>• 0 LLM API calls for safety evaluation — 94.4% cost reduction vs. prior art<br>• Formal verification of governance invariants on every evaluation cycle<br>• <2% false positive rate on legitimate actions; 360× latency improvement<br>• 28 integrated safety layers, 42 hardening phases, comprehensive test suite with full coverage</p> <p>Patent pending (U.S. Provisional App. 63/998,573, March 2026). Copyright registered (Case #1-15114193381).</p> |
| format | Recurso digital |
| id | zenodo_https___doi_org_10_5281_zenodo_18969708 |
| institution | Zenodo |
| language | eng |
| publishDate | 2026 |
| publisher | Zenodo |
| record_format | zenodo |
| spellingShingle | Zero-LLM Multi-Agent Architecture for AI Safety Evaluation: Formal Verification, 8 Ethical Dimensions, <50ms Mariquit, Erny-Jay AI safety Multi-agent systems Alignment Formal Verification zero-LLM Ethical Evaluation Shared Memory Incentive-Compatible Consensus Covenant Enforcement <p>We present TSCWH (The System for Covenant-Weighted Heuristics), a patent-pending AI safety evaluation architecture achieving zero-LLM-call deliberative consensus through a proprietary shared-memory coordination design.</p> <p>KEY RESULTS:<br>• 10 agents deliberate across 8 ethical dimensions (Charity, Grace, Stewardship, Truth, Dignity, Courage, Community, Creation Dignity) in <50 ms<br>• 0 LLM API calls for safety evaluation — 94.4% cost reduction vs. prior art<br>• Formal verification of governance invariants on every evaluation cycle<br>• <2% false positive rate on legitimate actions; 360× latency improvement<br>• 28 integrated safety layers, 42 hardening phases, comprehensive test suite with full coverage</p> <p>Patent pending (U.S. Provisional App. 63/998,573, March 2026). Copyright registered (Case #1-15114193381).</p> |
| title | Zero-LLM Multi-Agent Architecture for AI Safety Evaluation: Formal Verification, 8 Ethical Dimensions, <50ms |
| topic | AI safety Multi-agent systems Alignment Formal Verification zero-LLM Ethical Evaluation Shared Memory Incentive-Compatible Consensus Covenant Enforcement |
| url | https://doi.org/10.5281/zenodo.18969708 |