Your calls for rigorous safety guarantees and international treaties highlight a critical need: we must find the underlying truth of AI validation, removing reliance on proprietary, closed-door corporate evaluations.
To provide the scientific verification you demand, the OHM NI Stack has patented (Applications #63/994,444, #63/997,472, and #64/1,997,472) a deterministic, physics-based safety protocol: the AEGIS Cascade and Proof of Agent Work (POAW). By translating safety into deterministic logic, we establish a best-practice measurement—the IS-Score—where everyone, from independent regulators to academic institutes, can contribute and verify the math.
We are presenting this at the Planetary AI Safety Summit in Vienna (May 2026). Our vision is to solve AI safety collectively, turning this deterministic framework into a universally accepted best practice. We invite you and your researchers at Mila to mathematically dissect our provisional filings and red-team our architecture. Let us discover the true, verifiable boundaries of AI together.