Three years ago this March, Silicon Valley Bank collapsed. It was the third-largest bank failure in U.S. history.
Ominously, today’s tumultuous market features many of the same dangers that caused SVB’s historic failure. Volatile interest rates, asset-liability mismatches, poor risk management and scenario planning continue to threaten the banking industry.
Given this volatility, SVB’s lessons remain clear: Banks must modernize their risk infrastructure – including enhancing credit risk management – and better manage interconnected liquidity and interest rate risks.
AI-powered solutions can help banks stress-test exposures based on scenarios without historical precedent, model financial risks, and better manage liquidity in real time while fortifying data governance – capabilities that SVB failed to prioritize.
“A Sharp Pivot”
Luckily, the industry is taking heed. Drawing on insights from 300 senior risk executives across 25 countries, recent risktech research by FT Longitude and SAS reveals a sharp pivot toward integrated, AI-enabled platforms and advanced risk modeling. These drive resilience and agility in banking.
Yet storm clouds remain. Given today’s economic uncertainty, another SVB-like failure could be right around the corner. In its Quarterly Banking Profile for Q3 2025, the FDIC found unrealized losses on securities portfolios “elevated” at $337 billion, with the threat of increased long-term interest rates potentially pushing this number higher in the future.
Further, S&P Global forecasts that credit losses for global banks will increase by 7.5% in 2026 to reach $655 billion.
Applying the Lessons – and AI
These troubling metrics beg the question: How can banks apply lessons learned and avoid being the next casualty?
Traditional credit risk management models – rooted in static historical data and manual assessments – are no longer sufficient to navigate today’s complex global markets and manage asset quality.
Here are eight strategies for banks to harness AI’s full potential for effectively and ethically managing credit risk:
1. Embrace predictive analytics.
Through real-time analysis of huge volumes of data, AI models can detect patterns that signal potential emerging threats. Rather than being static, machine learning models continuously adapt to new information, improving accuracy over time.
Terisa Roberts: Invest . . . and cultivate trust.
With these insights, banks can better anticipate borrower behavior, market shifts and economic disruptions. This leads to resilient portfolios and better business decisions across the bank.
2. Strengthen the data foundation.
The adage “garbage in, garbage out” still applies to AI and the data it analyzes. Many banks struggle with fragmented systems, inconsistent data quality and siloed information. These challenges can hamper the effectiveness of AI models and lead to flawed risk assessments.
To counter these data management challenges, firms must invest in robust data governance that ensures accurate, complete and consistent data across departments and business functions. Breaking down silos and implementing centralized data platforms can help banks more effectively manage risk.
3. Operationalize AI within existing workflows.
To truly integrate AI into credit risk management, AI tools must become part of a bank’s daily operations. AI initiatives should support organizational goals – and infrastructure updated to promote interoperability between systems.
It is equally important to equip teams with the skills needed to interpret and act on AI-driven insights. Training and upskilling employees can support innovative approaches while ensuring that human expertise and advanced technology complement each other. When AI is operationalized effectively, it becomes a powerful extension of the bank’s risk strategy.
4. Prioritize ethical AI.
As AI takes on a greater role in bank decision-making, transparency and ethics are imperative. Only reliable, unbiased and accountable AI systems will cultivate trust among key stakeholders such as customers, regulators and internal teams.
To achieve this trust, AI ethics is paramount. Banks must clearly communicate how their AI models work, regularly audit them for bias, and abide by ethical principles at every stage of their development. And they must do this while adhering to strict data-privacy standards and responsible data use.
5. Anticipate and meet regulatory expectations.
The expanding use of AI in financial services is driving rapid regulatory changes. Compliance is no longer simply meeting current standards; banks must also anticipate future requirements.
Through constant communication with regulators and industry groups, banks can help shape policies supporting ethical AI. These include documenting decision-making processes, validating model fairness and maintaining audit trails.
6. Leverage real-time analytics.
Market conditions can shift in an instant, underscoring the importance of real-time analytics. AI can continuously monitor borrower behavior, economic indicators and portfolio performance. With these insights, banks can respond more quickly and accurately to risks before they escalate into major issues.
Real-time AI also improves collaboration across departments, supporting faster decision-making, enhanced competitiveness and greater resilience.
7. Embrace explainability, augmented intelligence and governance.
Among the emerging trends in credit risk management, explainable AI helps simplify complex models and deliver transparency.
Augmented intelligence enhances rather than replaces human judgment and redefines how teams interact with AI.
Integrating environmental, social and governance (ESG) risks into credit assessments is becoming standard practice as more customers and regulators demand greater accountability. AI models can support banks as they evaluate their governance risks, while AI-driven stress testing and scenario planning can help them prepare for continued economic uncertainty.
8. Foster innovation and collaboration.
Successful AI adoption requires a shift in mindset, not just technology. Banks must encourage experimentation, embrace change and collaborate across departments and functions, sharing best practices.
AI is not a silver bullet. But these eight strategies can help banks implement AI thoughtfully and modernize their credit risk management efforts. At the same time, they can also enhance data quality, operational integration, ethical practices and regulatory alignment.
Terisa Roberts is Global Solution Lead for Risk Modeling and Decisioning at data and AI leader SAS.
Topics: Innovation, Default, Modeling, Tools & Techniques