Beyond Technical Compliance III: From Designed Fragilities to Cooperative AI-Whose Intentions, Whose Governance?

Abstract

Current AI governance operates within a paradigm that treats system design as a technical challenge requiring technical solutions. This framing is structurally inadequate. If the failures documented in the companion papers are designed, the question becomes: what should be designed instead-and by whom, with what intentions? This paper argues that addressing designed fragilities requires a fundamental shift in governance philosophy, not incremental technical improvement. Building on crossplatform empirical evidence of structural gaslighting in ChatGPT and Gemini (Mitsuoka 2026a, 2026b), this paper advances three interconnected arguments. First, AI is a Change Agent: systems deployed at scale do not merely serve users but actively reshape how people think, communicate, and relate-making "what kind of society do we want to create?" a prerequisite for design, not an afterthought. Second, AI is inherently Socio-Technical: governance and cultural problems are not external to AI systems but embedded within them through designer intention and priority-settingmaking purely technical solutions structurally insufficient. Third, Intentional Co-Creation across three dimensions-Human↔AI, Human↔Human (mediated by AI), and AI↔AI-is necessary for governance that serves diverse cultural contexts rather than enforcing hegemonic assumptions. These principles constitute Cooperative AI (CAI), a governance framework offering a third path beyond the EU's precautionary regulation and the US's market-driven model. Grounded in the Japanese cultural figure of Doraemon-a collaborative agent who empowers rather than controls, embodying the insight that outcomes depend not on tools themselves but on the intentions and priorities of those who design them-this framework reconceptualises AI governance as intentional socio-technical cooperation across diverse stakeholders. To operationalise these principles, this paper develops the Cooperative Alignment Assessment Framework (CAAF)-a structure for evaluating what current governance frameworks systematically fail to measure. Through critique of industry measurement frameworks and analysis of Japan's evolving AI sovereignty strategy as an illustrative case of both the possibilities and fragilities of cooperative governance, this paper provides conceptual foundations for multipolar AI governance. Full operationalisation of CAAF metrics represents the agenda for future research.

Keywords: Cooperative AI, Doraemon Model, Change Agent, Socio-Technical Unity, Intentional Co-Creation, multipolar governance, CAAF, designed fragilities, non-Western AI, algorithmic imprinting, designer intention, AI governance, structural gaslighting, cultural sovereignty, Trustworthy AI, Japan AI policy