Lloyd Blankfein just put his finger on why even Goldman Sachs is wary of AI agents
Fortune • • business
Key Points:
- Lloyd Blankfein highlights that the primary risk of AI in finance is not superintelligence or autonomous weapons, but the inability to verify AI outputs for accuracy, which is critical in high-stakes institutional settings.
- Financial institutions fear rapid, unchecked AI-driven transactions because errors can cascade quickly, as demonstrated by past incidents like the 2010 flash crash and the 2012 Knight Capital loss, both caused by algorithmic failures.
- A 2026 Deloitte analysis identified over 350 risks from autonomous AI behaviors in banking, emphasizing that current regulatory and governance frameworks lag behind AI deployment, raising concerns about accountability and trust.
- Surveys show widespread use of AI in finance despite low trust in its accuracy; 97% of firms maintain human oversight, and many have encountered AI hallucinations or errors, underscoring the challenge of explainability and monitoring.
- Goldman Sachs exemplifies cautious AI adoption by running legacy and new systems in parallel and enforcing human approval for significant transactions, contrasting with faster AI deployment at other firms that may lack rigorous stress testing.