Generative AI can sharply raise efficiency and accuracy in corporate finance and tax operations, cutting routine costs and freeing staff for higher‑value work; but realising these gains at scale hinges on addressing data privacy, model reliability, system integration and regulatory accountability.

Main Finding

The paper finds that applying generative AI within corporate financial sharing centers (illustrated by Xiaomi’s Financial Sharing Center) and professional services firms (Deloitte) materially improves the efficiency and accuracy of finance and tax operations — notably in accounting, fund management, tax filing, and risk control — while also lowering costs and supporting decision-making. However, deployment faces important technical, governance, and organizational challenges (data privacy, model reliability, integration, regulatory/compliance and talent gaps), and these must be addressed to realize broad, safe adoption.

Key Points

• Scope of application
  • Accounting automation: automated bookkeeping, reconciliations, journal entry suggestion and error detection using LLMs and document understanding.
  • Fund management: cashflow forecasting, anomaly detection, and automated workflows for payments and collections.
  • Tax declaration: extraction of tax-relevant facts from invoices and contracts, draft tax returns, compliance checks, and scenario simulations.
  • Risk control and audit: real-time monitoring, fraud detection, KYC/AML screening, and automated exception reporting.
• Technical characteristics leveraged
  • Large language models (LLMs) for natural language understanding, drafting, and question-answering.
  • OCR + structured information extraction for invoices, receipts, contracts.
  • Retrieval-augmented generation (RAG) and knowledge bases for grounding outputs in firm data and regulations.
  • Process automation / RPA for integrating model outputs into workflows.
• Benefits observed
  • Increased processing speed and throughput for routine finance/tax tasks.
  • Higher consistency and reduced human error in repetitive tasks.
  • Cost savings through labor reallocation and task automation.
  • Better decision support via scenario analysis and anomaly prioritization.
• Challenges and limitations
  • Data privacy, confidentiality, and cross-border data transfer concerns.
  • Model hallucinations, lack of explainability, and limited audit trails.
  • Integration complexity with legacy ERP/financial systems and existing sharing-center processes.
  • Regulatory uncertainty around AI-generated filings and responsibility/liability.
  • Human capital: need for new skills, governance roles, and change management.
• Recommendations (high level)
  • Combine AI with human-in-the-loop controls and clear escalation paths.
  • Build governance, explainability, and auditability into deployments.
  • Start with pilots on high-volume, well-structured tasks; iterate and scale.
  • Invest in secure, high-quality training datasets and process integration.

Data & Methods

• Research objects: Xiaomi’s Financial Sharing Center and Deloitte (case examples).
• Methods:
  • Case analysis: in-depth qualitative examination of how generative AI is used in the two organizations’ finance/tax operations.
  • Literature review: synthesis of existing research and industry reports on generative AI capabilities and risks in corporate finance and taxation.
  • Analytical framing: mapping generative AI technical features to finance/tax tasks and evaluating impacts on efficiency, cost, and risk.
• Evidence type: qualitative, case-driven insights rather than new large-scale quantitative measurement. Findings are illustrative and context-dependent; generalizability may be limited by case selection and lack of standardized metrics.

Implications for AI Economics

• Productivity and cost structure
  • Generative AI can raise labor productivity in finance/tax, shifting work from routine processing to oversight, exceptions handling, and higher-value analysis.
  • Unit costs for bookkeeping and compliance tasks are likely to fall, affecting pricing in professional services and possibly leading to consolidation.
• Labor markets and skills
  • Demand will grow for hybrid roles: finance professionals with AI literacy, data governance, model validation, and control expertise.
  • Potential displacement of lower-skill transaction-processing roles; need for upskilling and redeployment policies.
• Market structure and competition
  • Firms that successfully integrate AI into centralized financial sharing centers can achieve scale economies, raising entry barriers for smaller competitors.
  • Professional service firms may shift from labor-intensive models toward productized AI-enabled offerings.
• Tax bases, compliance, and public policy
  • Automated tax-preparation and filing could increase compliance rates but also make tax bases more sensitive to automated tax-optimization strategies, requiring updated regulatory oversight and audit tools.
  • Cross-border data flows and AI decisioning will create jurisdictional and liability complexities for tax administration.
• Measurement and externalities
  • Productivity gains may be under- or mis-measured if GDP/tax systems aren’t adjusted for AI-driven quality changes in services.
  • Systemic risk: widespread deployment of similar models could create correlated failures or fraud vectors; macroprudential attention may be warranted.
• Policy and governance implications
  • Need for standards on explainability, audit trails, and certification for finance/tax AI systems.
  • Data governance frameworks and secure, privacy-preserving model training are essential.
  • Public–private coordination to update regulatory guidance on AI-generated filings, liability, and audit procedures.

Overall, the cases show generative AI’s potential to transform corporate finance and taxation by boosting efficiency and enabling new capabilities, but economic benefits will depend on addressing governance, regulatory, integration, and workforce challenges.