The Growing Concerns Around AI in Finance
Lloyd Blankfein, who spent decades at Goldman Sachs, has seen the financial industry weather numerous crises. From the 1987 stock market crash to the 2008 financial crisis, he has witnessed how risk management is crucial in such a volatile environment. His recent comments on the potential risks of artificial intelligence (AI) have sparked significant interest, especially given his extensive experience in managing large-scale operations.
Blankfein’s concerns are not centered around the idea of superintelligent machines or autonomous weapons. Instead, he highlights a more pressing issue: the inability to test whether AI systems are correct. In a world where precision is paramount, this lack of verification poses a serious challenge. When running a major institution, even small errors can have significant consequences, making it imperative to ensure that AI systems operate accurately and reliably.
Speed Without Oversight: A New Risk
The financial industry has long understood that speed can be a powerful tool. However, this leverage can quickly turn into a liability if not properly managed. A well-timed trade can lead to substantial gains, but a single mistake executed at machine speed can result in massive losses. This is precisely what Blankfein is warning about.
Historical events such as the 2010 “flash crash” and the 2012 Knight Capital disaster serve as cautionary tales. These incidents demonstrated the dangers of algorithmic trading and software glitches, which can lead to significant financial losses in a matter of minutes. With the new generation of AI agents being faster and more autonomous, the risks associated with their deployment are even more pronounced.
A March 2026 Deloitte analysis highlighted over 350 distinct risks that can arise from autonomous or agentic behavior in banking alone. Many of these risks are not addressed by existing frameworks, emphasizing the need for improved oversight and regulation.
The Data Behind the Concerns
Data supports Blankfein’s instincts regarding the limitations of AI in finance. A January 2026 Wakefield Research study found that only 14% of CFOs fully trust AI to deliver accurate accounting data without human intervention. Despite this, most firms are already using AI tools, with 97% of respondents stating that human oversight remains critical for accuracy.
The CFA Institute’s 2025 report on explainable AI in finance pointed out the challenges of transparency in AI-driven systems. A separate LinkedIn analysis from January 2026 further emphasized that supervisors often lack consistent, granular data on how AI is being used, and existing model risk management frameworks struggle with traditional validation, monitoring, and auditability.
The Race to Deploy AI
Despite these concerns, the deployment of AI in finance is accelerating rapidly. Ninety-two percent of leading fintech firms had integrated at least one autonomous agent into core production as of Q1 2026. However, governance frameworks are struggling to keep pace with this rapid development.
Goldman Sachs, JPMorgan, and Citi are among the major players investing heavily in AI. Goldman has rolled out its AI assistant to all 46,000-plus employees, while JPMorgan has over 450 AI use cases in production. Citi has more than 70% of its 182,000 employees using firm-approved AI tools. Yet, nearly all have drawn the same line: autonomous execution above certain thresholds still requires human sign-off.
The Importance of Caution
Blankfein’s approach to system transitions emphasizes running legacy and new systems in parallel for years before making a full switch. This method ensures that any issues can be identified and addressed before they become problematic. It is a discipline that many technology companies do not share, and it is increasingly at odds with the “move fast” culture that defines the current AI deployment wave.
The implicit warning is clear: firms that aggressively deploy AI agents may not have adequately stress-tested what happens when those agents make mistakes. As the industry continues to race toward AI integration, the need for careful oversight and robust governance becomes ever more critical.
