Generative AI moves into core finance workflows

Generative AI is being used across finance operations in banks and non-financial companies, from automating accounting tasks such as auditing, invoice capture and accounts payable to generating investment summaries, loan applications, client reports and regulatory submissions. Financial institutions are also deploying conversational systems that provide customer support, personalized financial advice, payment notifications and document generation, while Morgan Stanley uses OpenAI-powered chatbots to help financial advisors draw on internal research and data.

Back-office applications include modernizing legacy code, converting older languages such as COBOL, Fortran and other systems into modern languages while preserving business logic and improving documentation. Goldman Sachs has confirmed generative AI is part of its application development and enhancement work, and technology costs make up ~10% of a typical bank’s expenses, making faster development and lower maintenance costs financially meaningful.

Risk, markets and compliance teams are using generative AI for forecasting, scenario analysis, portfolio management, fraud detection, synthetic data generation, regulator responses and sentiment analysis. Mastercard used generative AI to scan transaction data across millions of merchants, doubling its detection rate for compromised cards, reducing false positives by up to 200% and increasing merchant fraud detection speed by 300%.

Key risks include poor data quality, inherited bias, limited performance from off-the-shelf large language models, hallucinations, regulatory pressure and data security concerns. Spending expectations remain high: by 2030, the banking industry is expected to spend 84.99 billion US dollars on generative artificial intelligence (AI), growing at a compound annual growth rate of 55.55 percent.