AI, crypto, and shadow banks are quietly reshaping global financial risk

The global financial system is undergoing a transformation unlike any before. But as innovation accelerates, safeguards have not kept pace. Fledgling artificial intelligence (AI) tools, unregulated crypto assets, and the expansion of shadow banking are quietly introducing new systemic risks. These threats are decentralized, opaque, and difficult to isolate—far more complex than the interconnected subprime mortgage crisis that set off the 2008 meltdown.

What makes these risks especially dangerous is their invisibility. They don’t sit on a single balance sheet or concentrate in a single geography. They are embedded in algorithms, distributed ledgers, and lightly regulated private lending networks. While these tools are often marketed as optimizing efficiency, they may be amplifying instability.

London School of Economics professor Jon Danielsson argues that AI is not just a neutral tool. “The problems originate from how algorithms collect and analyse data before executing solutions,” he says. These systems lack judgment, context, and caution—three traits that matter deeply when navigating turbulent markets.

Meanwhile, global leaders are distracted. Wars in Eastern Europe and the Middle East, tariff standoffs, inflation battles, and electoral cycles all crowd the geopolitical calendar. With major economies still growing and stock markets remaining resilient, there is little urgency for reform. But complacency may prove to be the biggest risk of all.

AI is already embedded in the financial system. From algorithmic trading and credit scoring to portfolio optimization and fraud detection, machine learning models are making decisions at speed and scale. But most of these systems operate as black boxes: even their creators often can’t explain how they arrive at decisions.

This becomes dangerous in moments of stress. AI models trained on historical data can misread unprecedented situations. They are prone to overfitting, can be manipulated through data poisoning, and often fail to adapt to nonlinear market behavior. A minor signal picked up by multiple models could be interpreted as a major trend, triggering simultaneous trades across firms.

The result? Herd behavior at machine speed—without the human pause for second-guessing. And unlike in 2008, when regulators at least understood how mortgage-backed securities worked, few central banks today have the technical capacity to audit neural networks running across proprietary hedge fund systems. Danielsson warns of an “arms race of blind automation,” where firms race to outdo one another in model sophistication, while regulators lag behind. The illusion of control may be worse than no control at all. Without interpretability, oversight becomes performative, not protective.

Cryptocurrency markets have quieted down since their 2021 boom, but the risks have not gone away. If anything, they’ve become more structurally embedded. Institutional players are now involved. Pension funds and payment platforms are gaining exposure. Central banks are experimenting with digital currencies. The fringe has gone mainstream.

Yet crypto remains fundamentally unstable. Many tokens still operate in legal gray zones. Their valuations remain speculative. “Stablecoins,” often marketed as low-risk alternatives, depend entirely on the credibility and auditability of their backing—something rarely proven in real time. More concerning is the infrastructure role that crypto is beginning to play. Blockchain-based systems are now involved in settlement, cross-border payments, and decentralized finance (DeFi) applications. If a major protocol fails, the collateral damage could extend well beyond token holders.

DeFi platforms, in particular, present a new challenge. These peer-to-peer systems bypass traditional banks and operate autonomously via smart contracts. When hacks or bugs occur—as they have repeatedly—billions in value can vanish overnight. And because DeFi is permissionless and global, it’s nearly impossible to enforce jurisdiction or restitution.

Shadow banking—a term for non-bank financial institutions (NBFIs) like hedge funds, money market funds, and private credit—has exploded in recent years. According to the Financial Stability Board, these players now control over $60 trillion in assets. They provide credit, liquidity, and investment services that rival traditional banks, but without the same regulatory oversight.

Why does this matter? Because shadow banks are deeply embedded in global funding markets. They borrow short-term to lend long-term, rely on repo markets, and engage in highly leveraged strategies. In a crisis, their flight from risk could drain liquidity from the entire system.

Complicating matters, data on these entities is limited. Few are required to report leverage or stress test results. When trouble starts brewing, regulators may have no idea where the cracks are until it’s too late. In 2023, the IMF issued a warning about rising leverage in private credit markets, but meaningful action hasn’t followed. As investors chase higher returns, many are funnelling capital into private debt funds without fully understanding the risk. These funds often lend to mid-market firms with little collateral, and they’re increasingly financing leveraged buyouts, real estate, and riskier business expansions.

For businesses: The growing complexity and opacity in the financial system mean that risk is harder to assess and price. A sudden liquidity freeze, a smart contract failure, or a bad AI output could derail financial operations or credit lines. Businesses—especially those dependent on fast capital—should reassess their exposure and build in more buffers.

For investors: Markets look calm, but volatility is hiding in plain sight. Investors should consider the interconnected nature of today’s assets: how a crypto shock might affect equities, or how AI-driven trading could amplify swings. Relying on past diversification strategies may be inadequate. Risk modeling itself needs a rethink in the age of machine-led finance.

Regulators face a dilemma: act too early and risk stifling innovation, act too late and face systemic collapse. Global coordination is crucial. AI, crypto, and DeFi operate across borders. Regulation must move beyond patchwork efforts and instead align on data transparency, algorithmic accountability, and shadow banking disclosure.

AI’s use in financial regulation must also be accelerated. Rather than banning its use, regulators should develop in-house capabilities to understand, test, and audit these tools. A baseline standard for explainability and safety is essential—not just for fairness, but for system stability.

We’re standing on shifting ground. The illusion of safety—because markets are stable and growth continues—should not blind us to the fault lines developing below. The next crisis may not come from mortgages or sovereign debt, but from a rogue algorithm, a collapsed DeFi platform, or a liquidity panic in private credit. Regulators must stop playing catch-up. AI, crypto, and shadow banking aren’t niche side stories—they are the new architecture of global finance. Without proactive guardrails, we risk stumbling into a breakdown no single institution or government can contain.

We don’t need to fear technology, but we do need to govern it. The time for pilot programs, fragmented oversight, and moral hazard is over. The lesson from 2008 was clear: complexity without transparency breeds disaster. Let’s not wait to relearn it.