Apollo Global Management's chief economist Torsten Slok raised an alarm about how artificial intelligence agents could destabilize the banking system by automating deposit migration toward higher-yield products.
The threat centers on a simple behavioral shift powered by automation. Today, many households keep deposits in low-interest checking accounts despite access to better returns elsewhere. This inertia benefits banks that fund loans and operations with cheap deposits. Slok argues that AI agents, operating on behalf of retail customers, would eliminate this friction. Once programmed to maximize returns, these agents would automatically sweep cash into money market funds, high-yield savings accounts, or other competing products the moment rates shifted.
The consequence looks straightforward: mass deposit outflows that squeeze bank funding models. Banks rely on deposit spreads, the gap between what they pay depositors and charge borrowers. Low-cost deposits form the bedrock of that equation. If AI agents yank deposits at machine speed whenever better yields appear elsewhere, banks lose that stable, cheap funding base. The result resembles a bank run, except triggered not by panic but by algorithmic optimization.
This scenario connects directly to recent banking stress. The 2023 failures of Silicon Valley Bank and Signature Bank revealed how fast deposits can flee when customers see risk or opportunity. Both banks faced deposit instability that ultimately proved fatal. AI agents would weaponize that instability by removing human hesitation from the equation. No more waiting for customers to manually research alternatives or overcome switching costs. An AI agent simply moves money.
Slok's warning extends beyond deposits alone. He frames this as a structural threat to the banking model that has persisted since the 1930s. Banks have always competed for deposits, but they competed in an environment where switching costs were high and information asymmetry favored the institution. AI agents flatten both advantages. Switching becomes instantaneous. Better rates become transparent across all options simultaneously.
The crypto angle here deserves attention. Decentralized finance protocols already offer programmatic yield farming and liquidity pools that operate on similar principles. Stablecoins and on-chain alternatives remove the need for traditional bank deposits entirely. If AI agents increasingly route household capital into DeFi or crypto-native yield solutions, traditional banks face a two-front squeeze: direct deposit losses to competing fintech, and capital migration to decentralized alternatives.
Regulators probably need to address this before it accelerates. The Federal Reserve already manages deposit insurance and bank reserve requirements, but those tools were designed for human behavior and slower capital flows. AI agents operate at network speed. A coordinated shift of deposits driven by algorithmic optimization could happen in hours, not days.
Banks will respond by raising deposit rates to compete. Some already have. But a sustained arms race pushes up their funding costs and compresses margins. The cheaper deposit advantage disappears entirely if they must match every competitor's yield.
Slok's warning cuts to the heart of how technology rewires financial infrastructure. AI doesn't just make existing processes faster. It removes behavioral inefficiencies that entire business models depend on. For banking, that's the low-cost deposit base. For crypto, it's another narrative supporting decentralized alternatives to traditional finance.
