Artificial intelligence (AI) has captured the attention of CEOs, CFOs, and treasurers worldwide, unlocking possibilities that would have been difficult to imagine just a few years ago—while introducing a distinct set of risks. Used effectively, AI can accelerate growth, improve forecasting accuracy, enhance liquidity visibility, and automate routine processes. Used poorly—or adopted without sufficient controls—it can amplify fraud risk, undermine data integrity, expose organisations to privacy violations, and erode trust.

AI also remains the dominant growth narrative in global markets, attracting unprecedented investment and executive attention. Yet as adoption accelerates, so do the trade-offs. Finance and treasury executives increasingly face a dual mandate: extracting value from AI while ensuring governance, resilience, and control. This tension is particularly acute in corporate treasury, where payments, cash forecasting, fraud prevention, and data quality sit at the intersection of automation and risk.

Against this backdrop, this article distils six essential AI insights, drawn from credible sources, research findings, and trusted reporting, to help business leaders, finance chiefs, treasurers and their teams evaluate current approaches and operating practices and prepare for what lies ahead. The objective is not to restate external analysis, but to interpret them through a treasury and finance lens—connecting innovation to operational realities to support informed decision-making in 2026 and beyond.

Agentic commerce: the future is already here—and it will transform payments

Agentic commerce refers to transactions in which AI acts on behalf of consumers or businesses—sourcing, assessing, and completing purchases or commercial transactions using preapproved payment methods. Unlike traditional automation, agentic systems do not simply assist; they execute. In payments, this represents a shift from user-initiated checkout to AI-initiated transactions that operate largely in the background.

Industry observers argue this is not a distant scenario. Writing in The Next 2025: The Collection, Karen Webster, CEO of PYMNTS.com, notes that in agentic commerce, “the critical shift is from what to buy to how to pay,” with payment optimisation emerging as the true differentiator. In this model, checkout disappears, replaced by intelligent credentials and AI agents that select payment methods, maximise rewards, and complete transactions autonomously.

Webster further observes that “Agentic commerce is coming. Not eventually but now, even though it’s very early days.” She adds that agents are being built by major technology companies and platforms to act as intermediaries between consumers and merchants, evaluating options, comparing total costs, assessing merchant reliability and return policies, and completing purchases in milliseconds, often with minimal—or in some cases no—human intervention beyond the initial prompt.

Data from PYMNTS Intelligence suggests consumer readiness is already taking shape. Nearly 70% of consumers express interest in using AI agents to simplify shopping tasks, while 41% want an agent to find the best deal for every purchase. One in three say they would trust an agent to choose the merchant, and nearly half of Gen Z and millennials expect AI to make purchasing decisions for them within the next five years.

Momentum is also building at the ecosystem level. As recently reported by CNBC, US players such as OpenAI, Perplexity, and Amazon—alongside Chinese technology firms including Alibaba and ByteDance—are pushing to embed agentic capabilities into commerce and payments ecosystems.

For corporate treasurers, this points to a gradual but meaningful shift in how payments may be initiated, authorised, and monitored. As procurement, expense management, and subscription activity become more automated, agent-driven transactions introduce new considerations around control frameworks, payment visibility, and fraud oversight—particularly where existing approval models are designed primarily for human-initiated activity rather than agent-led decisioning.

Open AI’s cash burn and classic bubble signs for AI on display

While AI adoption continues to accelerate, the financial sustainability of the AI ecosystem itself is coming under closer scrutiny. Reporting by The Economist indicates that OpenAI expects to burn roughly US$17 billion in cash in 2026, up from $9 billion in 2025, with cumulative losses continuing to mount over the coming years. Despite raising more than $60 billion since late 2022, questions around long-term profitability remain unresolved.

These dynamics have revived broader concerns about a potential AI investment bubble. Writing in the Financial Times, Ruchir Sharma, chairman of Rockefeller International, argues that AI now displays all four classic bubble characteristics: overinvestment, overvaluation, over-ownership, and over-leverage. He points to Big Tech’s increasing reliance on debt to fund large-scale AI build-outs and estimates that a significant share of U.S. GDP growth in 2025 was driven by AI-related activity.

For corporate treasury teams, these dynamics matter. A sharp correction in AI-driven equity markets could affect capital-raising plans, credit conditions, and counterparty risk—particularly given the leverage involved. Treasury forecasting models may therefore need to account for both optimistic and adverse AI scenarios, ensuring liquidity planning remains resilient even if today’s AI growth narrative undergoes a recalibration in 2026.

The US, China, and Singapore lead in AI enterprise adoption

AI enterprise adoption is increasingly concentrated among a small group of countries, led by the United States, China, and Singapore. While AI usage has spread rapidly at the consumer level, enterprise deployment tells a more differentiated story shaped by talent availability, capital investment, infrastructure readiness, and regulatory frameworks. According to analysis shared by Matt Li, co-founder of Second Talent, leading AI indices assess adoption not simply by usage, but across multiple dimensions, including the percentage of enterprises running AI in production, levels of public and private investment, AI talent density, government initiatives, infrastructure readiness, and research output.

Synthesising data from Stanford’s AI Index, Tortoise Global AI Index, and IMD’s Digital Competitiveness Rankings, LI’s analysis shows that the United States maintains its position as the global leader in enterprise AI adoption, with more than 85% of enterprises deploying AI. This leadership is underpinned by massive venture capital investment, dominant technology firms, and the world’s largest concentration of AI researchers, with particular strength across finance, healthcare, and technology sectors. China follows closely, with enterprise adoption exceeding 80%, driven by large-scale government investment, extensive data availability, and rapid deployment of AI across manufacturing, e-commerce, surveillance, and fintech. Singapore ranks third globally, combining high enterprise adoption with strong government AI readiness, regulatory clarity, and a national AI strategy that positions the city-state as a regional hub for finance, logistics, and technology sectors.

Source: Second Talent, Stanford’s AI Index, the Tortoise Global AI Index, and IMD’s Digital Competitiveness Rankings.

Beyond the top three, countries such as the UK, Germany, Israel, South Korea, Canada, the UAE, and Japan form the next tier of AI enterprise adopters, each with distinct strengths ranging from industrial AI and robotics to cybersecurity and government-led deployment.

For multinational treasurers, this uneven adoption landscape complicates standardisation. Treasury teams increasingly operate across jurisdictions with differing AI maturity levels, regulatory expectations, and infrastructure capabilities—affecting cash forecasting accuracy, fraud detection effectiveness, and the scalability of AI-enabled treasury platforms. As AI becomes more embedded in treasury and finance operations, global treasury strategies will need to accommodate local AI realities rather than assume uniform enterprise readiness across markets.

AI-powered malicious chatbots are accelerating payments fraud

AI is not only empowering legitimate automation—it is also reshaping financial crime. Malicious tools such as FraudGPT and WormGPT are enabling cybercriminals to generate highly targeted phishing emails, vendor change requests, and social-engineering scripts with unprecedented realism. These tools lower the barrier to entry while significantly increasing the scale and sophistication of payments fraud attacks.

Unlike earlier business email compromise schemes, AI-powered fraud is hyper-personalised and harder to detect. Attackers are also using deepfakes and jailbroken chatbots to bypass onboarding and identity-verification controls. For treasury, accounts payable, and procurement teams, this represents a significant escalation in risk.

The implication is clear: traditional controls are no longer sufficient. Treasury teams must strengthen vendor onboarding, payment instruction change verification, segregation of duties, and real-time anomaly detection. Planning for more targeted—and more deceptive—payments fraud attempts is now a baseline risk assumption.

CFOs see AI as transformative—but remain wary of privacy and security

New research from Kyriba, based on responses from 1,400 CFOs across eight countries, reveals a defining paradox. While 67% of CFOs expect AI to be the biggest driver of finance transformation over the next five years, 77% cite privacy and security as critical risks. This trust gap is shaping how finance leaders approach AI deployment.

Monica Green Boydston, chief product officer at Kyriba, characterises this dynamic as central to how finance leaders are approaching AI today. “What we’re seeing in 2026 is a perfect example: CFOs are highly optimistic about AI’s transformative potential, yet they’re approaching implementation with strategic caution around security and privacy,” she observes.

In response, CFOs are prioritising efforts to narrow this gap in 2026, focusing on AI adoption (53%), data reliability (31%), and security and fraud prevention (27%). This signals a broader push to strengthen operational foundations, ensuring new technologies are deployed with both confidence and control.

Agentic AI hinges on data readiness and strong governance

Agentic AI represents a step beyond traditional automation. As outlined in Strategic Treasurer’s 2026 Treasury Technology Analyst Report, agentic AI is designed to understand user intent, formulate multi-step action plans, and dynamically adapt based on context.

In treasury, this could enable goal-oriented execution—resolving exceptions, reconciling transactions, or managing workflows with minimal human intervention. However, delivering on this promise hinges on data quality. Poorly structured historical data, system silos, and inconsistent access can quickly undermine agentic performance.

Equally critical is governance. Agentic systems require clearly defined boundaries: approval thresholds, escalation logic, audit trails, and compliance controls. Autonomy without guardrails introduces risk. For treasury teams, success lies in balancing innovation with discipline—ensuring AI operates as a controlled extension of policy rather than a substitute for it.

Conclusion

AI is reshaping treasury and finance faster than any prior technology wave—but speed does not eliminate responsibility. For treasurers, CFOs, and CEOs, the challenge in 2026 is not whether to adopt AI, but how to do so responsibly. Agentic commerce, AI-driven forecasting, and automation offer real advantages, yet they also demand prudent governance, better data, and heightened vigilance against payments fraud.

The six insights outlined here point to a common conclusion: AI’s value in treasury depends less on algorithms and more on foundations. Data readiness, well-defined control frameworks, and rigorous risk management will determine whether AI becomes a durable competitive advantage—or a costly source of exposure. As the next phase of AI adoption unfolds, disciplined execution will matter more than ambition alone.

 

 

 ⃰ Disclosure: Strategic Treasurer owns CTMfile.