The Ex-IBM Chief AI Officer Who Left in 2022 Says the Industry's AI Safety Model Is Structurally Broken.
His Lab Just Published the Alternative. Releasing the "Unfireable Safety Kernel" a shipped, verified enforcement layer that sits outside the AI agent it governs RIDEGFIELD, CT, UNITED STATES, July 22, 2026 /EINPresswire.com/ -- ARYA Labs, PBC, made publicly available the source code, adversarial test suite and formal-verification artifacts for the Unfireable Safety Kernel, an authorization system designed to sit outside an artificial-intelligence agent's own runtime and remain in force even …