Case Studies
I do not use client documents for case studies. Client work stays confidential. Instead, I select high-stakes public documents, subject them to the full adversarial pipeline, and publish the structural findings with DOIs. The work is the proof. Download the reports. Verify the quotes against primary sources. Test the method yourself.
What This Is
These are adversarial reviews of major public documents. Each case study shows the full red-team process: dozens of attack passes, independent role councils, multi-model stress testing, and a public verdict.
If your organization relies on a document that has never been attacked this way — a contract, a policy, a regulatory filing, an AI governance framework — then it has undiscovered structural weaknesses. The proof below demonstrates what happens when I find them.
This is not theoretical commentary. This is applied topological failure analysis.
Three-pass adversarial structural analysis of the EU AI Act's systemic-risk provisions using the full AION diagnostic pipeline. 18 structural failure nodes identified across three independent passes. The Unified Thesis: the EU regulated systemic cognitive infrastructure using product safety mechanics designed for discrete products — a category error that underpins every downstream failure.
Survived multiple rounds of external hostile review by a 15-role rejection council. Zero fabrication found. 12 quotes independently verified against primary sources. 6 external legislative claims confirmed.
Key Finding
"Do no harm" has no definition, no enforcement mechanism, and no adjudicating body. The most trusted promise in medicine is structurally hollow — and AI governance frameworks are now being tethered to it.
Independent adversarial structural review of medicine's foundational ethical texts. "Do no harm" is undefined, unenforceable, and has no adjudication mechanism or technological update trigger. The review found 21 structural failure nodes across the Oath and the Declaration of Geneva, then proposed a Hardened 2026 Reconstruction with six-category harm definitions, independent counter-expert review, and AI-surgical oversight.
Anchored to existing regulatory frameworks — EU AI Act, MDR, GDPR/HIPAA — rather than inventing new abstractions. Every limitation named. Every claim cited.
Independent adversarial structural review of the ISA's draft Regional Environmental Management Plan guidance, run while the document was in active negotiation at the United Nations. 52 structural failure nodes found across four sections. Four critical gaps left the primary environmental protection obligation legally unenforceable from adoption.
Every finding traced to the document's own provisions. No invented mechanics. The primary finding includes the proposed resolution architecture as a proof of method.