Anthropic Co-founder Proposes 'Test Once, Mutual Recognition' AI Regulation at IMF Meeting

At the International Monetary Fund (IMF) and World Bank annual meetings held in October 2026, AI dominated discussions among global policymakers and financial leaders. During a roundtable titled "AI and the Global Economy: From Risk to Opportunity," Anthropic co-founder Jack Clark warned about the potential dangers of rapid AI development, comparing it to "an F1 car driving at high speed at night without brakes."

Clark noted that AI is subject to less regulation than baby toys. He pointed out that even plastic toys for toddlers have ingredient labels, toxicity testing standards, and mutual recognition mechanisms across countries, while AI oversight remains far weaker. He proposed a cross-border regulatory framework based on "test once, mutually recognized" to avoid inefficiency caused by duplicative oversight.

IMF Managing Director Kristalina Georgieva acknowledged at the forum that AI is spreading at an unprecedented pace, but the gains are concentrated in a few countries, and rapid technological change poses challenges to financial stability. She suggested that major AI developers and producers take the lead in bringing top tech companies to the table, working with industry self-regulation and public-private cooperation to define safety boundaries.

On employment, Georgieva cited IMF research showing that about 10% of jobs requiring new AI skills enjoy a significant wage premium, while routine and administrative roles face replacement risks. Low wages in services also create a mismatch between workers' salaries and career aspirations. Ayesha Khanna, co-founder of Addo, recommended drawing lessons from Singapore, offering free skills training and transition subsidies to enterprises. Bank of Canada Governor Tiff Macklem stressed that businesses must share the responsibility of transition with governments and retrain workers internally, otherwise society may face long-term backlash.