TL;DR:
- Ripple CTO Emeritus David Schwartz revisited Gary Gensler’s 2020 AI paper, agreeing with much of its concern about financial risk.
- The paper warned that widespread deep-learning adoption could make markets more fragile if institutions rely on similar models, data and optimization strategies.
- Schwartz questioned whether smarter systems would make irrational decisions, shifting attention toward correlated behavior, incentives and how autonomous agents could affect financial stability at scale.
Ripple CTO Emeritus David Schwartz has weighed in on a resurfaced 2020 paper co-authored by former SEC Chair Gary Gensler on AI and financial stability. Schwartz said much of the argument made sense, while questioning the idea that capable systems would become smart enough to behave irrationally. His response revives a debate over whether widespread AI adoption could create financial risks through synchronized decision-making. The discussion arrives as AI agents interact with financial infrastructure rather than remaining limited to research.
I think a lot of this makes sense. The part that doesn't is the "they'll be so smart that they'll do dumb things" part.
— David 'JoelKatz' Schwartz (@JoelKatz) September 28, 2026
AI Agents Revive Gensler’s Financial Stability Warning
Gensler co-authored the paper, titled “Deep Learning and Financial Stability,” with Lily Bailey in November 2020, before becoming SEC chair. The research argued that deep learning adoption across finance could create fragility if institutions relied on similar models, data and optimization strategies. The concern was less about individual algorithms failing and more about many systems reacting similarly at the same time. That possibility has become more relevant as AI agents move toward autonomous financial activity, including investing, payments and portfolio management.
The discussion also referenced concerns that autonomous agents could rapidly shift deposits or investments when optimizing returns, amplifying stress during market disruptions. Schwartz appeared to accept the coordination risk while rejecting the idea that greater intelligence naturally leads systems toward poor decisions. His distinction focuses attention on incentives and correlated behavior rather than assuming smarter models will simply act irrationally. That question matters as programmable payment systems give software the ability to execute financial actions continuously at machine speed.
The 2020 paper also argued that financial regulation, designed before widespread deep-learning adoption, might struggle with risks created by interconnected automated systems. The issue becomes one of market structure when similar models optimize against the same signals and react faster than human decision-makers can intervene. Financial institutions and crypto networks are experimenting with autonomous software, including XRP Ledger tools for AI-driven payments, making the earlier warning easier to connect with infrastructure.
Schwartz’s reaction does not endorse every conclusion in the paper, but it acknowledges that parts of the argument remain credible. The unresolved question is whether AI makes finance more efficient without making collective behavior more synchronized and fragile. As autonomous agents gain access to wallets, payment rails and investment tools, that balance between individual optimization and system-wide stability is moving from theory toward an operational concern today.
