Jane Fraser doesn’t mince words. The Citi chief executive sees artificial intelligence forcing banks into two contests at once. One pushes for growth. The other guards against collapse.
“There are two races in AI at the moment,” Fraser told the South China Morning Post. “One is to apply AI to the business models, which we all have to do as that will help drive revenue growth.”
She spoke days ago. Her comments landed amid fresh data showing banks view AI as both golden ticket and loaded gun. Surveys from early 2026 already echo her framing. They reveal executives chasing gains while bracing for smarter attacks.
Short product cycles. Fresh revenue streams. Smoother service. These mark the offensive push. Fraser highlighted how AI trims development time and boosts efficiency. It creates products once unimaginable. In corporate banking, tools shine at handling customer data and complex offerings, per longstanding McKinsey analysis referenced in coverage of her remarks.
Retail and private banking stand primed for agentic advisers. These systems grasp client needs, manage interactions, and execute. The Next Web noted this potential in its report on Fraser’s interview. Banks that deploy such agents first gain edge. They serve customers faster. They cross-sell smarter.
Yet speed brings risk. The second contest demands banks pour resources into shields. Fraser described it plainly. “These AI models are very powerful, and our job is to make sure the financial system, the bank, our customers and the ecosystem we operate in are secure.” Citi invests heavily. The goal? Keep fraud, money laundering and cyber threats from taking root.
Deepfakes already test trust. Voice cloning fuels social engineering. The CSI Banking Priorities Executive Report from January 2026 found AI-enhanced social engineering the top cybersecurity worry among 252 financial leaders. It jumped 16 percentage points. At the same time, 57 percent see AI itself as the best defense.
Numbers tell a tale of tension. Eighty-five percent of those surveyed believe AI adopters win competitive advantage. Half called it the leading technology trend. Banks showed twice the interest of credit unions in using it for back-office efficiency. Opportunity sits beside peril. Nancy Langer, cited in the CSI report, called AI both “path forward” and “new category of risk.”
The Revenue Chase Gains Speed
Fraser’s bank raised its sights. In March it lifted global AI capital expenditure and revenue forecasts for 2026-2030 to $8.9 trillion from $8 trillion. Faster enterprise uptake and agentic workflows drove the revision. The move signals confidence. It also shows how quickly expectations shift.
Industry forecasts paint upside. SAS experts predicted in late 2025 that retail banks could lift noninterest income 20-30 percent through commerce media models fed by verified financial data. Agentic AI moves from pilots to production. Semiautonomous systems handle real customer requests. They orchestrate workflows. They explain decisions.
“According to IDC, financial services firms will spend more than $67 billion on AI by 2028,” noted Diana Rothfuss, global solutions strategy director at SAS, in the SAS press release on 2026 predictions. “Production deployments… poised to see the biggest growth.”
Generative AI tackles unstructured data that makes up over 80 percent of enterprise holdings. It extracts insight at scale. Terisa Roberts, SAS global director for risk modeling, said it “will become for unstructured data what traditional statistics has long been.” Banks gain sharper risk views. They spot patterns earlier.
But adoption gaps yawn. A Forbes council piece from June warned most banks lag. Only 19 percent of institutions had moved agentic AI into production despite 96 percent running experiments. The gap between leaders and others widens fast. Winners redesign workflows around the technology. They don’t bolt it onto old processes.
Citi cuts illustrate the human side. The bank shed 3,500 technology roles in China last year. Those positions supported Asia-Pacific operations from Shanghai and Dalian. Fraser accepts change. “There will be job dislocations,” she said. “The nature of many jobs is going to change.”
She drew from memory. As a young analyst she photocopied microfiche and faxed documents to New York. That work vanished. New tasks appeared. “I really do see a lot of AI augmenting human beings, but there will be some dislocations as well.” The transition won’t align perfectly. “Our people are very adaptable.”
Defensive Lines Harden as Threats Multiply
Offense alone invites trouble. Fraser stressed parallel effort. Banks must outrun threats that grow more sophisticated by the month. Criminals wield the same models. Romance scams now deploy large language models for mass emotional manipulation. Deepfakes blur real from fake at scale.
SAS forecasts warn of synthetic data leaking into core repositories. It introduces bias and distortion into credit, fraud and risk models. Banks race to vault trusted data sources and limit generative AI contact with them. Ian Holmes, SAS director for enterprise fraud solutions, called it “a new kind of data integrity crisis.”
Agentic commerce adds fresh exposure. Autonomous agents may make unauthorized buys. Disputes follow. Hijacked agents trigger fraud. Banks need authentication for these digital actors through tokens or behavioral markers. Adam Neiberg at SAS expects “a surge in disputes triggered by autonomous AI agents.”
Trust emerges as measurable. Alex Kwiatkowski, SAS director of global financial services, put it starkly. “AI has made financial institutions faster, smarter and infinitely more confident – sometimes too confident… Trust must be earned, not assumed. In 2026, trust will morph from a promise to a performance metric.”
Regulators watch. The Cambridge Centre for Alternative Finance’s 2026 Global AI in Financial Services Report notes financial firms outpace rule makers in adoption. Fintechs lead incumbents. Competition and consolidation loom by 2030. Those who balance both races position best.
Fraser’s twin contests capture the moment. Banks cannot choose one. Grow too slow and rivals overtake. Protect too narrowly and threats breach. The finish line rewards those who advance revenue while fortifying foundations. Citi’s own forecast bump and job adjustments show it acts on both fronts.
Other voices align. A PwC analysis suggested full AI embrace could improve efficiency ratios by 15 percentage points. Accenture pointed to potential 30 percent productivity gains for institutions that redesign around the technology. Yet hype mixes with caution. Smaller players may make poor credit calls amid excitement. Fraser has flagged such exuberance before.
The stakes rise daily. Synthetic data contaminates. Agents act independently. Fraudsters scale scams. Banks that treat the two contests as separate lose ground. Those who run them in tandem write the next chapter of finance. Fraser made the choice clear. Compete on growth. Defend the system. Do both. Relentlessly.
WebProNews is an iEntry Publication