The writer, chair and CEO of Goldman Sachs, argues that AI regulation must be tailored to the technology, not copied from banking. He highlights the Mayo Clinic’s AI model detecting pancreatic cancer early, showing AI’s transformative potential. While optimism about AI’s benefits is warranted, he warns against a false binary of unchecked progress versus heavy restrictions.
He outlines three layers of responsibility: governments setting rules, developers building safe models, and companies deploying AI — the least discussed but most critical layer. Grafting bank regulations onto AI oversight is the wrong approach. Instead, deployers must adopt principles like accountability, counterparty risk management, stress testing, and encouraging risk escalation.
Every company is now an AI company, and managing operational risk is essential. AI systems must be independently reviewed, monitored, and stress-tested. The goal is to advance AI responsibly, with tailored scrutiny that matches its pace and complexity.
Source: Financial Times Companies · Summarized by HeadlinesBriefing