Martin Wolf argues the AI surge threatens personal accountability, the rule of law and even the definition of humanity. He cites a recent Anthropic blog saying AI systems are increasingly designing their own successors and that slowing progress would be wise. Competition between US firms and China fuels rapid development, making effective regulation extremely difficult and could reshape global markets overnight.

The International Labour Organization estimates one‑quarter of workers face some generative‑AI exposure, yet only 3.3 % sit in the highest‑risk bracket. Venture capitalist Vinod Khosla predicts AI will perform 80 per cent of economically valuable tasks across 80 per cent of jobs, implying massive under‑employment unless policy frameworks arrive swiftly. Past tech waves, such as electricity, showed long lags before productivity gains materialised and could reshape labour negotiations.

Wolf calls for an internationally binding AI testing and liability regime, with the EU positioned as a potential regulator of last resort. He warns that unchecked AI could concentrate economic power, widen inequality and even enable autocratic threats to democracy. Immediate action means governments must codify criminal and civil penalties for AI‑induced harm across all sectors, by law today.