The Hidden AI Risk Emerging From CBDC Adoption

Central bank digital currencies (CBDCs) are seen as a key financial innovation of the digital era that speeds payments, enhances transparency, improves financial inclusion, and strengthens monetary sovereignty. These capabilities are why CBDCs have attracted growing interest from central banks worldwide.

But what’s less discussed is the less visible risk of these digital versions of official currencies. The adoption of CBDCs is changing the way companies think about and present their technology strategies, and not always for the better.

Recent research suggests that CBDCs increase pressure to demonstrate digital transformation, which causes some companies to intensify their AI narratives without matching those claims with technical development. It’s more marketing than machine learning.

This gap between what firms say about their AI ambitions and what they actually build has real consequences for investors, capital markets, and the credibility of the broader AI investment cycle, and it has largely gone unexamined until now.

For investors, the lesson goes beyond CBDCs. Instead of relying only on management commentary, they must evaluate AI exposure through patents, R&D intensity, hiring, deployments, and revenue generation. As AI claims become ubiquitous, the credibility gap between promotion and execution may become a material valuation and financing risk.

Why Central Banks Are Racing Toward Digital Currencies

Most countries have their own central banks. These centralized authorities manage a nation’s money and credit supply through monetary policy. They utilize tools like interest rates, reserve requirements, and open market operations to control inflation, stabilize the value of money, and guide economic growth. While designed to operate independently, central banks do not really function¹ as such.

Not only is their independence in question, but these public financial institutions also face limits on monetary tightening and concerns about maintaining monetary sovereignty in the digital economy.

Over the past decade, financial systems have rapidly digitized and private payment platforms have grown. Today, payment systems, financial products, and forms of money are evolving faster than regulatory and policy frameworks, while stablecoins and tokenized assets operate alongside traditional structures, further weakening the impact of central banks’ policy decisions.

CBDCs have emerged in response to all these issues. While the idea of a central bank issuing its own digital money is not new, it gained significant traction after the rise of Bitcoin (BTC ), which was launched in the aftermath of the 2008 financial crisis that eroded public trust in financial institutions.

The rise of cryptocurrencies got the attention of monetary authorities and forced them to consider what state-backed digital money looks like. Over the last few years, central banks have accelerated research, experimentation, and pilot programs to modernize payment infrastructure while maintaining complete control.

Data shows significant interest in CBDCs, with 146 countries and currency unions involved in one way or another. Of these, 40 are currently researching the idea of creating digital versions of their national currencies. Even more countries (41) have already launched pilots, while many (33) are in the development phase. About fifteen such projects, however, are inactive and nine have been canceled.

Notably, only a handful of economies, like Nigeria, have gone all the way to a fully launched retail CBDC, but their experience has been one of modest adoption.

The Central Bank of Nigeria (CBN) recently shared its plans to use eNaira for government payments such as pensions, civil servant salaries, and social-welfare payments to boost adoption of the eNaira, which was launched in 2021, per the bank’s ‘Nigerian Payments System Vision 2028.’

Member nations of the BRICS group, which include Brazil, Russia, India, China, and South Africa, along with Nigeria among others, are also discussing linking their fast payment systems with CBDCs, Reuters reported.

“Cross-border payments is an area of interest for all of us, including the BRICS, because we feel there is a lot of scope for reducing cost. Various options are on the table, but it is still at the discussion stage, including CBDCs and linkages of fast payment systems.”

– Reserve Bank of India (RBI) Governor Sanjay Malhotra last week at an event

Besides internationalizing the local currency, the governor also said the RBI doesn’t see AI as a risk to be contained but as a capability to be harnessed.

“Indian banks cannot afford to sit on the sidelines and watch,” said Malhotra. His call to lenders to establish board-approved AI governance policies comes as central banks globally scrutinize lenders’ usage of AI amid worries over operational risks, governance risks, and cyberattacks.

“Innovation and safety are not opposing goals, they are in fact complementary requirements of a durable financial system.”

– Malhotra

In the realm of CBDCs, China’s digital yuan (e-CNY) initiative is one of the most prominent examples. The People’s Bank of China began internal CBDC research in 2014, launched a pilot program in 2019, and expanded it across major economic regions by the end of 2022.

e-CNY particularly stands out for its adoption. The CBDC has already processed over 3.4 billion retail transactions worth around 16.7 trillion renminbi (about $2.3 trillion) as of December 2025.

Earlier this year, the PBOC reclassified digital yuan as a deposit liability, signaling its evolution from a mere cash substitute into a more deeply embedded part of China’s financial infrastructure. The appeal of CBDCs lies in their ability to offer faster and cheaper settlement than traditional means like bank transfers.

Besides facilitating instant settlement, digital currencies also reduce transaction friction, lower payment costs, improve financial system efficiency and monetary policy transmission, and enhance transparency through more traceable transactions.

Enhanced traceability, in particular, is important to regulators, as it is what makes CBDCs so different from cash or even ordinary bank deposits. Unlike cash, CBDCs can create transaction records within a centrally administered payment system, potentially giving monetary and regulatory authorities greater visibility into how funds move.

This combination of features creates an information environment for corporate behavior that has not existed before, and it appears to be quietly reshaping how firms communicate about their technology strategies.

If companies become more visible to regulators, investors, and the public all at once, it raises the incentive for them to be seen doing the “right” things, even when the substance is simply not there. These micro-level consequences for businesses haven’t received much attention from researchers, who have largely focused on macroeconomic outcomes like monetary transmission, financial stability, and payment system reform.

Emerging evidence suggests that CBDCs do more than transform payment systems; they also transform the institutional environment in which firms operate. Increased visibility of financial activities, stronger expectations around digital transformation, and greater regulatory oversight may affect how companies communicate their innovation strategies and allocate resources. This creates a new type of risk around the authenticity of corporate AI investment.

This overlooked risk underpins a recent study examining the relationship between CBDC adoption and enterprise AI strategies, using China’s e-CNY rollout as a natural experiment.

How CBDCs May Be Fueling an AI Credibility Problem

The study titled All sizzle and no steak: The predicament of enterprise AI strategies under central bank digital currency policies² investigates whether CBDC adoption influences the gap between what firms say about AI and their actual AI investment. This gap is termed “AI strategy decoupling,” defined as the divergence between a firm’s public AI narratives and its real AI technological capabilities.

The researchers measured AI narratives through related disclosures in annual reports, while actual, verifiable capability was proxied using related patent applications drawn from the China National Intellectual Property Administration’s patent classification system.

Using a sample of more than 30,000 firm-year observations from 5,015 Chinese A-share listed companies between 2014 and 2024, authors Jinlei Yu and Rui Chen from the Department of Management, Business School, Nankai University, China, and Haotian Luo from the Department of Economics, City St George’s, University of London, UK, apply a staggered difference-in-differences design that exploits the fact that different Chinese cities entered the e-CNY pilot at different times.

This design allowed the researchers to estimate the effect of e-CNY adoption while accounting for broader changes affecting Chinese firms.

The e-CNY pilot program was used as a quasi-natural experiment because its staggered, multi-level launch across cities provides plausibly exogenous temporal and geographic variation for causal identification. The main finding is that CBDC adoption significantly increases AI strategy decoupling.

The e-CNY rollout significantly increased firms’ AI-related strategic communication while producing no statistically significant increase in AI patent applications, which the researchers used as a proxy for technological capability. Rather than accelerating genuine innovation, the policy environment only encourages symbolic compliance.

Firms signal alignment with digital transformation priorities without producing a corresponding increase in observable AI patent activity.

In this study, “we reconceptualize AI strategy decoupling as an institutionally conditioned response to specific domain characteristics rather than a product of managerial short-termism, deliberate machine-washing, or governance failure,” noted the authors. “By showing that decoupling arises systematically when high signaling value, large cost asymmetry, and low external verifiability co-occur under institutional pressure, we transform it from an ex post descriptive label into an ex ante predictive construct applicable to future technology domains and infrastructure reforms.”

Now, decoupling happens when companies adopt public commitments or formal policies that fit within institutional expectations while having internal practices that deviate from those positions.

In the context of CBDCs, this means firms embrace AI publicly in response to amplified institutional visibility but selectively adjust internal resource allocation away from building genuine AI capability. “This is not mere managerial negligence but a structured organizational response: firms actively manage the boundary between external-facing narratives and internal resource decisions to satisfy legitimacy demands at minimal cost,” says the study.

The researchers have identified three main mechanisms driving the gap. The first is enhanced capital liquidity, which lowers financing frictions. This makes symbolic signaling cheaper than committing capital to long-term AI R&D, meaning firms find it cheaper to talk than to build.

The second mechanism is greater transaction traceability, which increases the visibility of firms’ capital allocation decisions to regulators and investors. This incentivizes more narrative signaling while discouraging risky investment failures that would now be more visible than before.

“CBDC is an externally imposed institutional change that restructures the payment environment in which firms operate, altering financing conditions, monitoring intensity, and settlement speed,” noted the study.

Per institutional theory, such environmental shifts produce “coercive, mimetic, and normative pressures that compel organizations to conform to prevailing institutional expectations.”

So, when CBDC policy raises the visibility of corporate financial behavior and signals the state’s commitment to digital transformation, firms face greater legitimacy pressure to align with tech policy priorities.

The third mechanism is faster settlement speeds, which shorten the time horizon over which firms are judged. This rewards visible, frequent, and low-cost signaling over slow-maturing innovation projects.

Together, these three mechanisms increase the attractiveness of AI rhetoric while discouraging costly and uncertain investment in genuine AI capability.

Mediation analysis produced statistically significant evidence supporting all three proposed channels, and the results were robust to several checks, including parallel-trends tests, Goodman-Bacon decomposition, placebo tests, Heckman correction for selection bias, propensity score matching, and a stacked difference-in-differences design.

But why is AI particularly vulnerable to this kind of decoupling? That’s because it offers exceptionally high signaling value, requires substantial long-term investment to build actual capabilities like original algorithms and large-scale model training, and remains difficult for external stakeholders to verify.

“Under information asymmetry between firms and external stakeholders, firms have incentives to choose the conformity mode that maximizes reputational returns per unit of cost,” said the authors.

As the study noted, AI capability is opaque because it is embedded in intangible assets like algorithms and data pipelines.

Furthermore, “AI occupies the apex of China’s technology policy hierarchy and commands disproportionate investor attention and policy legitimacy, making AI-related claims far more rewarding than equivalent claims about cloud computing or big data.”

Notably, the decoupling effect is not evenly distributed. It concentrates among firms with higher existing digital transformation, operating in high-technology sectors where AI claims carry the most reputational weight, and located in regions where local government policy documents emphasize digital transformation most heavily. These firms also tend to have lower institutional investor ownership, meaning sophisticated shareholders are less present to discipline empty claims.

This means the firms most likely to talk big about AI without the capability to match are exactly those already embedded in digital and policy ecosystems that reward the appearance of innovation. Markets do not ignore this behavior entirely.

As per the study findings, firms with greater AI decoupling see lower valuations, weaker venture capital inflows, reduced profitability, and higher debt financing costs.

This suggests that creditors, equity investors, and venture capitalists can at least partially identify when AI claims are not supported by meaningful technological development and penalize symbolic AI signaling accordingly. But the fact that decoupling persists despite these penalties suggests the short-term legitimacy gains from AI talk still outweigh the long-term market discipline against it, for now.

The authors therefore highlight this as an unintended and previously undocumented consequence of CBDC adoption: while digital currencies can improve financial efficiency, they may distort corporate innovation disclosures well beyond the financial system they were designed to modernize, weakening the credibility of AI-related disclosures in the process.

International Business Machines Corporation (NYSE: IBM)

For investment purposes, IBM makes an attractive choice given its involvement in both domains: CBDCs and artificial intelligence. The company provides enterprise-grade blockchain frameworks and research tools to help central banks develop, test, and scale secure CBDCs. More recently, IBM launched Digital Asset Haven, a platform for financial institutions, governments, and corporations to securely manage and scale their digital asset operations. IBM’s established financial-services and payment-infrastructure business further gives it credible exposure to the modernization of regulated financial systems.

IBM was not examined in the study, and its inclusion here reflects its exposure to the technologies and governance needs highlighted by the research rather than any direct connection to the researchers’ findings.

Meanwhile, its enterprise-ready AI and data platform, watsonx, is designed to build, scale, and govern generative AI and machine learning workloads.

IBM is a $223.4 billion market cap company whose shares, at the time of writing, are trading at $236, up 11.34% in the past month but down 20% YTD and 2.22% over the past year. It has an EPS (TTM) of 12.45 and a P/E (TTM) of 19.05. IBM pays a dividend yield of 2.85%.

On the financials side, IBM reported revenue of $17.16 billion and adjusted EPS of $2.93 for Q2 2026. Revenue grew 1% YoY. Net income was $2.17 billion. By segment, Software revenue was up 5% to $7.8 billion. Hybrid cloud (Red Hat), automation, and data all increased, while transaction processing software fell 8%. Consulting revenues were flat at $5.3 billion, the Financing segment jumped 12% to $0.2 billion, and Infrastructure revenue fell 7% to $3.8 billion due to worse-than-planned Z mainframe computer sales, which were down 42%.

“We are taking action to accelerate our revenue growth and profitability, driving productivity across the company with AI and automation, and heavily investing in commercializing innovation at speed and scale.”

– CEO Arvind Krishna

IBM’s free cash flow in Q2 was $2.5 billion, down $0.3 billion YoY, and the company ended the quarter with $8.2 billion in cash, restricted cash, and marketable securities, down $6.3 billion from year-end 2025.

During the period, the company returned $1.6 billion to shareholders in dividends. For the full year, the company expects constant currency revenue growth of 4%-5% and free cash flow to increase by about $1 billion.

Going forward, IBM will focus on its high-growth portfolio, including software that helps clients manage, deploy, and build AI-ready solutions, where “client demand is strongest.” Quantum is another area of focus, with plans to invest over $10 billion. IBM also said it is on track to deliver the first large-scale fault-tolerant quantum computer by 2029.

“Although we faced revenue headwinds late in the second quarter, we continued to focus on the fundamentals of our business, including driving productivity, strengthening our portfolio, and generating free cash flow.”

– CFO James Kavanaugh

Conclusion

CBDCs are gaining traction among governments for good reason; they modernize payment systems, improve efficiency, and strengthen financial infrastructure.

But it’s not all gains; CBDC adoption also carries risks. Evidence suggests their impact goes beyond finance and into corporate innovation behavior, with the latest study finding that CBDC adoption unintentionally encourages firms to increase AI narratives without producing a corresponding increase in observable AI patent activity.

This creates a need for policymakers, technology providers, and investors to remain attentive to the quality of corporate AI disclosures. They must ensure that AI narratives match actual technological progress in order to fully realize the benefits of both CBDC and AI.

References

1. Eijffinger, S. & de Haan, J. Central bank independence and accountability. Journal of International Money and Finance, 166, 103595 (2026). https://doi.org/10.1016/j.jimonfin.2026.103595
2. Yu, J., Chen, R. & Luo, H. All sizzle and no steak: The predicament of enterprise AI strategies under central bank digital currency policies. International Review of Economics & Finance, 111, 105689 (2026). https://doi.org/10.1016/j.iref.2026.105689