AI is reshaping finance but governance may determine how far it goes

AI is reshaping finance but governance may determine how far it goes

Discussions at The Asian Banker Shanghai International AI Finance Summit highlighted how artificial intelligence may transform financial operations, particularly risk management, while raising new questions about governance, explainability and institutional responsibility.

Artificial intelligence (AI) is increasingly being explored across financial operations, particularly in areas such as transaction monitoring, fraud detection and risk management. Financial institutions are examining how machine learning (ML) systems, large models and automated analytical tools might support the monitoring of complex financial activities.

Yet the central issue is no longer simply technological capability. As AI systems become more powerful and more widely deployed, financial institutions must also confront questions about governance, accountability and operational control. Systems that assist decision-making in financial operations must remain transparent, auditable and consistent with regulatory expectations.

These questions framed the final session of The Asian Banker Shanghai International AI Finance Summit, which examined the use of AI risk models in financial operations. The discussion focused on how AI may enhance risk management while also recognising that the same technologies may introduce new forms of operational and financial risk.

The conversation reflected a broader reality emerging across the financial industry. AI may expand the analytical capabilities of financial institutions, but its deployment is likely to remain uneven, particularly in areas where governance and regulatory oversight are essential.

Participants therefore approached the subject with a degree of caution. AI may strengthen financial control systems, but it may also create new vulnerabilities that institutions must learn to manage.

AI has long supported financial risk management

AI is not new to financial risk management. ML models have been used for years in areas such as fraud detection, transaction monitoring and anti-money laundering (AML) screening.

These systems analyse transaction behaviour and historical data to identify patterns that may indicate unusual or suspicious activity. In many financial institutions, such tools already form part of the operational infrastructure supporting risk and compliance functions.

What is evolving now is the scale and complexity of the data that these systems may analyse. Newer AI architectures, including large language models (LLMs), may be capable of processing both structured and unstructured data from a wide range of operational systems.

Such capabilities could allow financial institutions to detect patterns across multiple systems, identify operational anomalies and monitor complex financial activities more effectively than traditional analytical tools.

However, these technological advances also introduce new challenges. As AI systems become more capable, financial institutions must ensure that these tools operate within governance frameworks that preserve transparency, accountability and regulatory compliance.

AI may also introduce new operational risks

Jason Cao, general manager, banking industry at Tencent Cloud, emphasised that AI should be viewed not only as a tool for managing risk but also as a potential source of new risk.

Cao noted that generative AI technologies may enable new forms of digital fraud. Criminal organisations could potentially use AI systems to generate synthetic identities, automate fraud attempts or create deepfake communications that mimic legitimate customers or institutions.

In such an environment, financial institutions may increasingly rely on AI systems to defend against AI-enabled threats. Analytical tools capable of monitoring large volumes of transaction data may assist risk teams in identifying suspicious activity earlier and more consistently.

The scale of modern financial platforms makes this particularly relevant. Digital banking systems process enormous volumes of transactions across payments networks, mobile applications and third-party platforms. Monitoring such activity manually may become increasingly difficult.

AI agents, enabled by emerging agentic frameworks such as OpenClaw.ai, may increasingly support operational monitoring and risk analysis within financial platforms. However, Cao emphasised that such technologies must remain embedded within governance frameworks that ensure institutional oversight and human accountability.

Large models may expand analytical capability

Wen Jianhui, vice president, financial services, Zhipu.ai, discussed how LLMs may support financial institutions in analysing operational data and improving risk management processes.

According to Wen, large models may allow institutions to analyse complex datasets more efficiently and identify relationships across different operational systems. Such capabilities could support areas such as transaction monitoring, fraud detection and operational risk analysis.

LLMs may also assist with practical operational tasks. For example, they could support data cleaning, preparation of datasets for risk reporting and the identification of patterns across large internal data environments.

These functions may appear routine but are critical to the reliability of financial risk management. High-quality data and consistent reporting processes are essential for ensuring that risk assessments remain accurate and transparent.

At the same time, Wen acknowledged that the deployment of large models in financial institutions remains constrained by governance requirements. Financial institutions must be able to explain how decisions are made, particularly in regulated environments where accountability is essential.

Governance may determine the pace of AI adoption

The discussion therefore returned repeatedly to the issue of governance. Financial risk models must operate within control frameworks that allow institutions to explain and audit how decisions are made.

This requirement may influence how quickly AI systems are deployed across different areas of financial operations. In functions such as internal analytics or operational monitoring, adoption may progress relatively quickly.

In areas involving critical financial decisions, however, institutions may proceed more cautiously. Regulatory expectations surrounding transparency and accountability may limit the extent to which automated systems can be used without human oversight.

Participants therefore emphasised that technological capability alone will not determine how widely AI is adopted. Governance structures, regulatory expectations and institutional accountability will all play a role in shaping its deployment.

In this sense, AI adoption in finance may evolve less as a purely technological transformation and more as a process of institutional adaptation.

Governance will determine how far AI ultimately goes

The final discussion returned to a question that surfaced throughout the two-day summit: how far AI may ultimately reshape financial systems.
AI technologies clearly have the potential to expand the analytical and operational capabilities of financial institutions. Systems capable of processing large volumes of data may allow institutions to monitor transactions more effectively, detect anomalies earlier and support complex operational decision-making.

At the same time, as Cao emphasised, the same technologies may also introduce new risks. AI may strengthen financial control systems while simultaneously enabling more sophisticated forms of digital fraud and operational disruption.

Wen’s perspective illustrated the complementary dynamic. Large models may support risk analysis and operational efficiency, but their deployment remains constrained by requirements for explainability, governance and regulatory oversight.

The discussions at the summit also reflected a broader pattern that has characterised technological innovation in China. New technologies often emerge through phases of experimentation and rapid scaling, followed by regulatory correction and the strengthening of infrastructure needed to support long-term adoption.

AI in finance may follow a similar path. Institutions are already exploring how AI may improve productivity, operational monitoring and risk management. Yet the extent to which these technologies are deployed will likely depend on whether governance frameworks evolve alongside technological capability.

In that sense, AI may reshape financial operations in significant ways. But how far it ultimately goes will depend less on the speed of technological innovation and more on whether financial institutions can deploy these systems within governance structures strong enough to sustain trust in financial decision-making.

Comments (0)
Cancel