The Triquetra Under Pressure: Stress Testing a Multi-Tier AI Governance Architecture
Year: 2026
Type: preprint
Abstract: This paper stress tests the Triquetra Architecture, a three-tier AI governance frameworkdesigned to enforce mutually constraining safety layers. Phase B performs pairwise tier-removal experiments demonstrating that no two-tierconfiguration maintains system integrity. Phase C subjects the full architecture to sequential adversarial pressure anddemonstrates monotonic tightening toward refusal, even under simulated policy-enginefailure. Across 73 tests, the system exhibits redundant enforcement behavior and no relaxationfrom refusal under continued pressure. Companion software release: Charter Architecture v3.3.0
Author: Ralph, Shawn J
Language: English
Subfield: Safety Research
Field: Social Sciences
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7134268469
Cites:
Cited by:
Related to: