AI is enhancing trading efficiency, ESG analytics and risk assessment in major financial centres but is also introducing opacity, bias and systemic fragilities; whether these benefits materialize for sustainable finance hinges on adaptive, coordinated regulatory oversight.

Main Finding

AI-driven innovation in financial markets (trading, risk management, compliance, sustainability analytics) can raise market efficiency, ESG integration, and risk‑assessment accuracy, but it simultaneously creates governance risks — notably model opacity, algorithmic bias, and possible amplification of systemic vulnerabilities. How these benefits and harms play out depends strongly on institutional governance: adaptive regulation, cross‑institutional coordination, and ethical oversight are critical to align AI adoption with long‑run financial stability and sustainability objectives. The study proposes an integrative sustainable‑AI governance framework to guide regulators and firms in capturing AI’s upside while managing systemic implications.

Key Points

• Positive effects documented
  • Improvements in market efficiency (faster processing, better price discovery).
  • Stronger ESG integration (AI helps scale sustainability analytics and integrate ESG signals into investment/risk decisions).
  • Enhanced risk assessment capabilities (richer models for credit, market, and operational risk).
• Governance challenges
  • Model opacity and explainability gaps that impede oversight.
  • Algorithmic bias and procyclicality risks that can produce unfair or destabilizing outcomes.
  • Potential amplification of systemic vulnerabilities (herding, feedback loops, correlated exposures).
• Conditioning role of institutions
  • Outcomes vary across regulatory environments; adaptive governance mitigates harms and helps translate AI capabilities into positive sustainability and stability outcomes.
• Policy orientation of the proposed framework
  • Emphasizes regulatory adaptability, institutional coordination (across domestic agencies and internationally), and ethical oversight (transparency, accountability) as core mechanisms.

Data & Methods

• Comparative mixed‑methods design covering leading financial systems (China, United States, United Kingdom) over 2022–2025.
• Comparative institutional analysis to assess how governance regimes shape AI adoption and effects.
• Data sources integrated:
  • Secondary quantitative indicators (e.g., market efficiency metrics, ESG integration measures, and risk indicators — used to evaluate observable market and risk outcomes).
  • Qualitative documentary evidence (regulatory texts, policy statements, industry reports and related documentary sources) to characterize governance approaches and practices.
• Synthesis approach: linking quantitative outcome patterns to qualitative evidence about governance to identify conditioning mechanisms and derive a governance framework.

Implications for AI Economics

• Modeling and measurement
  • Economic models should incorporate governance heterogeneity and algorithmic externalities (opacity, bias, feedback loops) when assessing AI’s effect on market efficiency and systemic risk.
  • Need for new metrics that capture AI‑specific systemic risk channels (e.g., algorithmic commonality, automated-herding measures, opacity-adjusted risk metrics).
• Policy and regulation
  • Adaptive, iterative regulatory regimes (rulemaking plus experimentation and rapid learning) are economically efficient for rapidly evolving AI technologies.
  • Macroprudential tools should be extended to account for AI‑driven correlation and amplification channels (stress tests, scenario analysis that include algorithmic behavior).
  • Disclosure, model documentation, and mandatory audits can reduce information asymmetries and lower regulatory frictions; these interventions affect compliance costs and market incentives and should be calibrated to avoid undue innovation costs.
• Market structure and competition
  • AI adoption can raise concentration and incumbency advantages (scale economies from data and models); antitrust and competition policy should be integrated into assessments of social welfare from AI in finance.
• Distributional and welfare considerations
  • Improved risk pricing and ESG integration can raise social welfare, but risks of bias and job displacement create distributional tradeoffs requiring complementary policies (retraining, fairness audits).
• International coordination
  • Cross‑border spillovers of algorithmic trading and globalized financial firms mean that international coordination reduces arbitrage and regulatory gaps; economics research should quantify benefits of harmonized standards.
• Research agenda
  • Empirical work to quantify net welfare impacts of AI in finance, decomposing efficiency gains from governance costs.
  • Design and evaluate market‑based incentives (insurance, liability rules) to internalize algorithmic externalities.
  • Develop causal evidence on how specific governance interventions (transparency requirements, audits, stress tests) change AI behavior and systemic outcomes.

Summary takeaway: AI materially reshapes financial‑market functioning and sustainability analytics, but the net economic impact hinges on governance design. AI economics should move beyond tech‑centric assessments and explicitly model institutions, regulatory adaptability, and coordination as core determinants of welfare and systemic risk.