THE next financial shock may not begin with a queue outside a bank and a nervous fund manager or a trader shouting “sell”.

It could begin quietly, when several computer models read the same piece of bad news, reach the same conclusion, and start moving money in the same direction before humans fully understand what is happening.

In a July 23 blog, International Monetary Fund (IMF) Financial Counsellor Tobias Adrian warned that artificial intelligence (AI) is becoming part of how financial institutions assess borrowers, price risk, and respond to market stress. That can make finance faster and more efficient. Under stress, however, it could make shocks spread more quickly and become harder to diagnose.

For Jamaica, the concern is no longer entirely theoretical.

CIBC Caribbean said in March that it was preparing to introduce lending solutions in Jamaica under which decisions on unsecured loans and credit cards would be driven by AI.

GK Capital has since launched an investment-research platform whose AI assistant answers questions about stocks, bonds and market developments. VM Investments Limited (VMIL) said last October that it was integrating AI into customer service and routine work supporting its research and risk-management teams.

The applications are not equivalent. CIBC would use AI in credit decisions. GK Capital’s assistant supplies information but does not execute trades. VMIL described the technology as supporting employees and operations.

None of this means Jamaica is approaching an AI-driven financial crisis. It means the technology is moving beyond experimentation and into services touching Jamaican borrowers and investors.

The question is whether the safeguards are developing just as quickly.

The same speed cuts both ways

AI could shorten loan approvals, identify suspicious transactions more quickly and give ordinary investors access to research previously used mainly by professionals.

CIBC said approximately 80 per cent of its clients across 10 Caribbean markets use digital channels, while 95 per cent of transactions are conducted electronically. The bank therefore already has the customer behaviour and technological foundation to automate more lending.

But speed, scale and automation are also at the centre of the IMF’s warning.

Systems used by different institutions may respond to the same market signals. During a crisis, several properly functioning models could therefore reach the same conclusion and adjust portfolios simultaneously.

Adrian said some AI-based funds already rebalance considerably faster than traditional strategies, potentially amplifying market swings.

Herd behaviour is not new. Investors have always followed one another and rushed towards the same exit during periods of fear. AI changes how quickly that can happen.

The Jamaican initiatives do not establish that computers are automatically trading securities on the Jamaica Stock Exchange. The warning is about where markets could be heading as AI becomes more deeply involved in financial decisions.

Who explains the decision?

Automated lending presents a more immediate issue for consumers, but Jamaican law already provides some protection.

Subject to exemptions, Section 12 of the Data Protection Act allows an individual to require that a significant decision not be based solely on automated processing used to assess matters including creditworthiness.

Where such a decision is made, the institution must inform the individual, who may request reconsideration or a new decision not based solely on automation. The information commissioner may intervene.

The Act contains exceptions, including for some contract-related decisions. The issue is how the right will work in practice and whether the reviewer has the authority and expertise to overturn the automated decision.

The risk hidden in shared suppliers

Financial institutions may appear separate while depending on the same cloud, data or AI providers. A failure or cyberattack affecting one supplier could disrupt several institutions at once.

VMIL’s overseas partnership does not establish supplier concentration in Jamaica. It illustrates why regulators need to know which critical services institutions obtain externally.

AI also strengthens attackers by producing more convincing phishing messages and helping fraud schemes adjust more rapidly.

Can the watchdogs keep up?

Bank of Jamaica (BOJ) has identified elements of the risk. Its 2024 Financial Stability Report described AI and cyber threats as growing concerns.

A November 2025 governance standard also says information-technology committees at deposit-taking institutions and financial holding companies should monitor AI and guide policies covering ethical use and data privacy.

BOJ is therefore not approaching AI from a standing start. Its published documents do not say whether institutions must report material AI use, whether it maps shared providers or whether stress tests examine several automated systems responding to the same event.

The Financial Services Commission faces similar questions because VMIL and GK Capital operate in the securities industry it supervises.

AI can help regulators analyse reports, transactions and complaints. But the IMF warns that the expertise required to examine complex systems is scarce, particularly in emerging markets. AI, it says, should augment supervisory judgement, not replace it.

The regulatory test is arriving while adoption can still be mapped and governed. Jamaica does not need to wait for a crisis to determine where AI is being used, which external systems institutions share, and who can intervene when automated decisions begin reinforcing one another.