From AI Pilot to P&L: A Board-Level Framework for AI Investment Accountability in ASEAN

Firms have adopted artificial intelligence (AI) widely, yet evidence of enterprise-level financial return remains thin. Large surveys in 2025 and 2026 find that most organisations cannot attribute material earnings impact to AI, and administrative data show little effect on earnings or hours despite reported time savings. This paper argues that, for many organisations, the problem facing boards is one of conversion and evidence rather than adoption: task-level gains reach the financial statements only when a financial mechanism is named in advance, freed capacity is converted into a measurable financial change, value is measured against a pre-committed baseline with an explicit attribution standard, and continuation is decided at staged gates. Drawing on benefits-realisation, IT-governance and real-options literatures, on contemporary AI value-measurement frameworks and on regulatory developments in Malaysia, Singapore and the wider Association of Southeast Asian Nations (ASEAN) as at 3 October 2026, the paper proposes the AI-to-P&L Accountability Framework: a seven-link evidence chain governed by non-compensatory gates, a conversion test, an attribution grade and a realised-value requirement, with AI risk clearance as a precondition. The framework assesses whether a value case is decision-grade; it does not replace the organisation's capital-allocation criteria. The paper sets out board reporting and disclosure principles, an analysis of how the framework can be gamed and testable propositions. It is a conceptual practitioner working paper, oriented to ASEAN through governance mapping rather than regional validation, and has not been peer reviewed.