Story audio is generated using AI
Imagine telling a personal finance assistant to protect your savings from a weakening currency. It monitors exchange rates, compares accounts and, with your permission, moves part of your balance into dollar stablecoins.
For you that might look like prudent housekeeping. Across millions of households, similar decisions could become a funding problem for domestic banks. This is a possible next chapter in dollarisation: software making currency substitution a routine financial habit.
The IMF has identified the underlying risks. In August remarks on emerging markets it warned that stablecoins can make foreign currency more accessible and that shifts toward reserve investments abroad can tighten domestic credit conditions. It also noted that AI could accelerate adoption.
The additional hypothesis is about speed and co-ordination. Households already seek dollars when inflation erodes savings or confidence weakens. An authorised agent could lower the effort involved, checking conditions repeatedly and acting within preset limits.
A banking system might consequently face persistent, automated withdrawals without a dramatic queue outside a branch. But the accounting matters.
Buying stablecoins does not automatically destroy bank deposits. If a customer buys tokens from another domestic resident the money may simply move between accounts. Issuer reserves may also remain in banks.
Bank for International Settlements analysis emphasises that the funding consequences depend on reserve composition and how money moves through the system. Retail deposits can become more concentrated wholesale funding, with different costs and stability characteristics.
The concern for emerging markets is sharper when transactions redirect funding abroad or shift deposits away from banks that cannot readily replace them.
Banks under pressure could raise deposit rates, seek alternative funding or adjust lending. Businesses and borrowers might then face higher costs or tighter access to credit, depending on the scale of outflows and policy responses.
A household could protect its purchasing power while its employer finds financing harder to obtain.
AI would not need to cause the original loss of confidence. It could accelerate responses to inflation, depreciation or unattractive deposit returns.
The potential feedback loop deserves attention: automated conversions increase demand for dollars, currency pressure strengthens the case for further conversions, and more agents respond.
That sequence is a scenario, not evidence that an AI-driven bank run is under way. Exchange controls, transaction costs, wallet access and customer permissions could all constrain it.
Nor are stablecoins equivalent to guaranteed dollar deposits. Their value depends on reserve quality, redemption arrangements and intermediaries. Conversion fees can erode savings. Products offering additional yield introduce risks beyond simply holding a dollar-linked token.
An agent comparing headline returns could overlook those differences unless its rules explicitly account for them.
There is also a co-ordination problem. Different assistants may use similar data, providers or financial thresholds. If many users authorise comparable responses, supposedly independent decisions could cluster around the same shock.
For policymakers, blocking a payment technology would not resolve the reasons people want another currency. More credible monetary policy, competitive savings products and dependable domestic payments could reduce that incentive.
The IMF also cautions that clear evidence of stablecoin-related banking disintermediation remains limited and that outcomes differ across countries.
The useful indicators would therefore be verified conversions from domestic deposits, the destination of corresponding funds, changes in bank funding costs and the role of automated instructions.
Token transaction volumes alone cannot establish capital flight. The predictive question is whether saving becomes an always-on optimisation process that domestic banks struggle to retain.
AI could improve household financial decisions while making deposits less predictable. A country’s next funding shock might begin with thousands of reasonable instructions: protect my savings, find a better return, move the money.
• Muchena is founder of Proudly Associated and author of ‘Artificial Intelligence Applied’ and ‘Tokenized Trillions’.
Business Day
