[論文レビュー] Carbon-Aware Governance Gates: An Architecture for Sustainable GenAI Development

Mateen Ahmed Abbasi, Tommi Mikkonen+4 more|arXiv (Cornell University)|Feb 23, 2026

Scientific Computing and Data Management被引用数 0

ひとこと要約

The paper proposes Carbon-Aware Governance Gates (CAGG), an architectural extension to GenAI governance that embeds emissions provenance, carbon budgeting, and green validation orchestration into SDLC governance layers to reduce the carbon footprint of AI-enabled development.

ABSTRACT

The rapid adoption of Generative AI (GenAI) in the software development life cycle (SDLC) increases computational demand, which can raise the carbon footprint of development activities. At the same time, organizations are increasingly embedding governance mechanisms into GenAI-assisted development to support trust, transparency, and accountability. However, these governance mechanisms introduce additional computational workloads, including repeated inference, regeneration cycles, and expanded validation pipelines, increasing energy use and the carbon footprint of GenAI-assisted development. This paper proposes Carbon-Aware Governance Gates (CAGG), an architectural extension that embeds carbon budgets, energy provenance, and sustainability-aware validation orchestration into human-AI governance layers. CAGG comprises three components: (i) an Energy and Carbon Provenance Ledger, (ii) a Carbon Budget Manager, and (iii) a Green Validation Orchestrator, operationalized through governance policies and reusable design patterns.

研究の動機と目的

• Motivate the need to address the environmental impact of GenAI-enabled software development and governance overhead.
• Propose a layered architecture that integrates sustainability into human–AI governance for GenAI-enabled SDLCs.
• Introduce three architectural extensions to governance: emissions provenance, carbon budgeting, and green validation orchestration.
• Define enforceable governance policies and reusable design patterns to operationalize sustainability in governance.
• Highlight trade-offs and limitations of integrating carbon-aware governance into existing DevOps ecosystems.

提案手法

• Present a carbon-aware governance reference architecture for GenAI-enabled software development.
• Describe three architectural extensions: Energy and Carbon Provenance Ledger, Carbon Budget Manager, and Green Validation Orchestrator.
• Define governance policies such as Model Escalation, Regeneration Cap Policy, Carbon-Intensity Scheduling, and Budget-Bound Validation.
• Propose four reusable architectural design patterns: Budgeted Governance Gate, Two-Phase Carbon-Aware Validation, Carbon Provenance Evidence, and Stop-and-Justify Regeneration Loop.
• Explain how the extensions enable a sustainability governance control loop with measurement, decision, and enforcement.

実験結果

リサーチクエスチョン

• RQ1 What architectural mechanisms and governance points can embed sustainability into GenAI-enabled software development?
• RQ2 How can emissions provenance, budgeting, and validation orchestration be integrated without restructuring existing governance control structures?
• RQ3 What policy and design patterns enable enforceable carbon-aware governance in SDLC workflows?
• RQ4 What trade-offs arise between assurance, performance, and carbon footprint in governance decisions?

主な発見

• CAGG embeds sustainability into governance by adding an energy and carbon provenance ledger, a carbon budget manager, and a green validation orchestrator.
• Governance policies (e.g., model escalation, regeneration limits, carbon-intensity scheduling, budget-bound validation) enable carbon-aware decision making in validation and oversight.
• Design patterns provide reusable solutions for budgeted governance, two-phase validation, provenance evidence, and stop-and-justify regeneration loops.
• The approach shifts sustainability from infrastructure optimization to architectural concern within GenAI governance, enabling explicit trade-offs to be managed at governance checkpoints.
• Acknowledges limitations such as carbon-footprint estimation accuracy, potential over-constraint in safety-critical contexts, and added architectural complexity.