CiberIA with AIsecTest and Ψ∑AISysIndex

This technical report introduces CiberIA, an innovative system designed to assess the functional security self-perception of artificial intelligences. At its core lies the AIsecTest, a 100-item evaluation instrument inspired by clinical and neuropsychological self-awareness scales, which measures an AI's ability to recognize its own limitations, detect errors, and assess internal vulnerabilities. The system applies a multi-judge scoring methodology, involving six AI models and one human evaluator, and translates the results into the Ψ∑AISysIndex, a standardized introspective maturity index. The report presents the conceptual foundation, system architecture, scoring methodology, and results of empirical simulations on large language models. It also discusses implications for AI auditing, regulatory compliance (such as with the EU AI Act), and the ethical evaluation of autonomous systems, proposing CiberIA as a foundational tool for next-generation AI governance.